Designing, Modeling and Control of a Tilting Rotor

Quadcopter

A Dissertation submitted to the

Graduate School
of the University of Cincinnati
in partial fulfillment of the
requirements of the degree of

Doctor of Philosophy

in the Department of Electrical Engineering and Computing Systems

of the College of Engineering and Applied Sciences
by

Alireza Nemati

M.S. Shahrood University of Technology

August 2007

Committee Chair: Manish Kumar, Ph.D.

Ali Minai, Ph.D.
 An Abstract of
Designing, Modeling and Control of a Tilting Rotor Quadcopter
by
Alireza Nemati

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the
Doctor of Philosophy Degree in Electrical Engineering
University of Cincinnati
March 2016

The aim of the present work is to model, design, control, fabricate and experimentally study

quadcopter with tilting propellers. A tilting quadcopter is an aerial vehicle whose rotors can

tilt along axes perpendicular to their respective axes of rotation. The tilting rotor quadcopter

provides the added advantage in terms of additional stable configurations, made possible by

additional actuated controls, as compared to a traditional quadcopter without titling rotors. The

tilting rotor quadcopter design is accomplished by using an additional motor for each rotor that

enables the rotor to rotate along the axis of the quadcopter arm.

Conventional quadcopters, due to limitation in mobility, belong to a class of under-actuated

robots which cannot achieve any arbitrary desired state or configuration. For example, the ve-

hicle cannot hover at a defined point at a tilted angle. It needs to be completely horizontal in

order to hover. An attempt to achieve any pitch or roll angle would result in forward (pitch) mo-

tion or lateral (roll) motion. This proposed tilting rotor concept turns the traditional quadcopter

into an over-actuated flying vehicle allowing us to have complete control over its position and

orientation.

In this work, a dynamic model of the tilting rotor quadcopter vehicle is derived for fly-

ing and hovering modes. The model includes the relationship between vehicle orientation an-

gles and rotor tilt-angles. Furthermore, linear and nonlinear controllers have been designed to

achieve the hovering and navigation capability while having any desired pitch and/or roll orien-

tation. In the linear approach, the four independent speeds of the propellers and their rotations

about the axes of quadcopter arms have been considered as inputs. In order to start tracking a

desired trajectory, first, hovering from the initial starting point is needed. Then, the orientation

ii
of the vehicle to the desired pitch or roll angle is obtained. Subsequently, any further change in

pitch or roll angles, obtained using a linear controller, result in motion of the quadcopter along

the desired trajectory.

The dissertation then presents a nonlinear strategy for controlling the motion of the quad-

copter. The overall control architecture is divided into two sub-controllers. The first controller

is based on the feedback linearization control derived from the dynamic model of the tilting

quadcopter. This controls the pitch, roll, and yaw motions required for movement along an ar-

bitrary trajectory in space. The second controller is based on two Proportional Derivative (PD)

controllers which are used to control the tilting of the quadcopter independently along the pitch

and the yaw directions respectively. The overall control enables the quadcopter to combine

tilting and movement along a desired trajectory simultaneously.

Furthermore, the stability and control of tilting-rotor quadcopter is presented upon failure

of one propeller during flight. On failure of one propeller, the quadcopter has a tendency of

spinning about the primary axis fixed to the vehicle as an outcome of the asymmetry about the

yaw axis. The tilting-rotor configuration is an over-actuated form of a traditional quadcopter

and it is capable of handling a propeller failure, thus making it robust in one propeller fail-

ure during the flight. The dynamics of the vehicle once the failure accrued is derived and a

controller is designed to achieve hovering and navigation capability.

The dynamic model and the controller of the vehicle were verified with the help of numer-

ical studies for different flight scenarios as well as failure mode. Subsequently, two different

models of the vehicle were designed, fabricated and tested. Experimental results have validated

the dynamical modeling and the flight controllers.

iii
 Copyright 2016, Alireza Nemati

This document is copyrighted material. Under copyright law, no parts of this document may
be reproduced without the expressed permission of the author.

iv
Acknowledgments

First and foremost, I would like to express my sincere gratitude to my advisor Professor

Manish Kumar for the continuous support of my Ph.D. study and related research, for his

patience, motivation, generosity, enthusiasm and immense knowledge. I appreciate all his con-

tribution of time, ideas, encouragement and financial support that he provided. His continuous

guidance helped me carry out research and write this thesis. During my study, he was not only

a great advisor for problems related to my work, but also an exceptional consultant for other

situations I faced.

I wish to express my sincere thanks to Professor Ali Minai, my thesis advisor, for his

valuable guidance and encouragement extended to me. I am thankful to Dr. Kelly Cohen for

all his motivations he gave not only to me but also to the other members of our team, for his

guidance and suggestions.

Besides my advisors, I would like to thank Professor Raj Bhatnagar, and Professor Rui Dai,

for their insightful comments, assistance and encouragement throughout my project.

I addition, a thank you to Younes Kheradmand, who was my boss before I started my Ph.D.

for 5 years. I really appreciate all the advise he provided me like a father and encouragements

he gave like a friend during my career. A thank you also to professor Abdullah afjeh who has

inspired me the professionalism.

My appreciation also extends to my fellow lab mates. Balaji R Sharma, Baisravan Hom-

chaudhuri and Ruoyu Tan who helped me settle in the lab for first couple of months. I would

like to thank my friends at the Cooperative Distributed Systems Laboratory for the wonderful

times, long meetings and exciting brainstorming discussions we had. Two of my colleagues

cannot remain nameless, Mohammad Sarim and Mohammadreza Radmanesh for the stimu-

lating discussions, for the sleepless nights we were working together before competitions and

deadlines, for the travels we made for different conferences and for all the fun we had in the
v
last couple years.

The final words in acknowledgments are usually reserved to those dearest to the author. I

do not wish to break this tradition. Without the love and support from my family, I would not

have come this far.

vi
Table of Contents

Abstract iii

Acknowledgments v

Contents vii

List of Tables x

List of Figures xi

List of Abbreviations xiv

1 Introduction 1

1.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.3 Motivation for this work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.4 Contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.5 Publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.6 Organization of Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2 History of the Quadcopters 9

2.1 A Brief History of Quadcopters . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.1.1 The Early History of Quadcopters . . . . . . . . . . . . . . . . . . . . 9

2.1.2 History of Tilt Rotor Vertical Takeoff Vehicles . . . . . . . . . . . . . 12

2.1.3 History of Tilt Rotor Quadcopter . . . . . . . . . . . . . . . . . . . . . 16

2.2 Current Quadcopters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.2.1 Quadcopters with Tilting Mechanism . . . . . . . . . . . . . . . . . . 20

vii
3 Dynamic Modeling 27

3.1 Traditional Quad-rotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.2 Tilting Rotor Quadcopters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

4 Control System 35

4.1 Linear Controller Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.1.1 Proportional Derivative Control . . . . . . . . . . . . . . . . . . . . . 35

4.2 Nonlinear Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.2.1 Feedback Linearization . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.2.2 PD Based Tilting Angle Controller . . . . . . . . . . . . . . . . . . . . 47

4.3 Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

5 Fault Tolerant Flight 54

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

5.2 Fault Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

5.3 Dynamic Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

5.3.1 Tilting Rotor Quadcopters with One Propeller Failure . . . . . . . . . . 57

5.4 Controller Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

5.5 Linearization and Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . 63

6 Hardware design 70

6.1 Description of the Prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

6.1.1 Central body . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

6.1.2 Tilting Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

6.2 Drive System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

6.2.1 Motors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

6.2.2 Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

6.3 Avionics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

6.3.1 Control Board . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

6.3.2 Communications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

viii
7 Numerical Simulations and Experimental Results 80

7.1 Numerical Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

7.1.1 Simulation set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

7.1.2 Tilt-Rotor Quadcopter Simulation results . . . . . . . . . . . . . . . . 81

7.1.3 Feedback Linearization Numerical Simulation . . . . . . . . . . . . . 84

7.1.4 Fault Tolerant Numerical Simulation . . . . . . . . . . . . . . . . . . . 87

7.1.4.1 Hover Flight . . . . . . . . . . . . . . . . . . . . . . . . . . 87

7.1.4.2 Tracking a Trajectory . . . . . . . . . . . . . . . . . . . . . 90

7.2 Preliminary Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . 94

7.2.1 Hovering on the spot . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

7.2.2 Trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

7.2.2.1 Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

7.2.2.2 Box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

8 Conclusions and future work 101

8.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

8.2 Future Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

References 104

ix
List of Tables

6.1 Specifications of the Prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

6.2 Vehicle’s component mass details. . . . . . . . . . . . . . . . . . . . . . . . . . . 74

6.3 Power output from ESC, 4S LiPo, 10x47 Propeller. . . . . . . . . . . . . . . . . . 78

x
List of Figures

2-1 Gyroplane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2-2 The Oemichen2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2-3 De Bothezat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2-4 Convertawings, Model A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2-5 George Lehbergers 1930 tilting propeller vertical take-off “flying machine”. . . . . 13

2-6 Three-view drawing of the Focke-Achgelis FA-269 convertiplane . . . . . . . . . 14

2-7 The Bell XV-3, during flight testing. . . . . . . . . . . . . . . . . . . . . . . . . . 15

2-8 XV-15 taking off . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2-9 The V-22 Osprey, during transition flight . . . . . . . . . . . . . . . . . . . . . . . 16

2-10 X-19 in hovering flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2-11 Bell X-22A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2-12 The Quad-TiltRotor concept, University of Patras. . . . . . . . . . . . . . . . . . 20

2-13 CAD design of the CQTR, Istanbul Commerce University. . . . . . . . . . . . . . 21

2-14 The Configuration of the QTR UAV under research a) Quadcopter mode b)Flying

wing mode, Beihang University. . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2-15 CQTR with integrated actuators in different flight configurations. . . . . . . . . . . 23

2-16 Overview of developed CQTR, Nihon University. . . . . . . . . . . . . . . . . . . 24

2-17 CAD model of the quadcopter with tilting propellers, Max Planck Institute. . . . . 25

2-18 Vehicle prototype on the ball joint rig flight test, Cranfield University. . . . . . . . 26

3-1 Schematic diagram showing the coordinate systems and forces acting on the quad-

rotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3-2 Coordinate Frames and Free body diagram of Tilting Quadcopter . . . . . . . . . 31

4-1 The block diagram of position and orientation control algorithm . . . . . . . . . . 36

xi
4-2 Hovering with tilted arms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

5-1 Free body diagram of tilt-rotor quadcopter upon propeller failure . . . . . . . . . . 58

5-2 Vehicle’s Position . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

5-3 Vehicle’s Orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

6-1 The CAD model of the tilting quadcopter . . . . . . . . . . . . . . . . . . . . . . 70

6-2 HS-5087MH HV Digital Micro Servo. . . . . . . . . . . . . . . . . . . . . . . . 72

6-3 The CAD model of components of the tilting mechanism. . . . . . . . . . . . . . 72

6-4 The real model of the first model of tilting Quadcopter. . . . . . . . . . . . . . . . 73

6-5 The CAD model of transparent tilting mechanism . . . . . . . . . . . . . . . . . . 75

6-6 The CAD model of tilting mechanism mounted on the arm with the servo and the

motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

6-7 The CAD and the actual model of the prototype. . . . . . . . . . . . . . . . . . . 76

6-8 850 kv brushless DC motors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

7-1 The reference (commanded) pitch and roll angle . . . . . . . . . . . . . . . . . . . 81

7-2 The actual trajectory followed by the UAV in 3-dimensions . . . . . . . . . . . . . 81

7-3 The actual orientation of the vehicle in 3 directions . . . . . . . . . . . . . . . . . 82

7-4 The angle of each arm during simulation . . . . . . . . . . . . . . . . . . . . . . . 82

7-5 The speed of each rotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

7-6 The enlarged view of the figure 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

7-7 Quadcopter trajectory in three-dimensional space . . . . . . . . . . . . . . . . . . 85

7-8 Position and orientation of the quadcopter: altitude vs. time (top left), x-position

vs. y-position (top right), pitch vs. time (bottom left), and yaw vs. time (bottom

right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

7-9 The reference roll (bottom) and actual roll angle (top) during the flight. . . . . . . 86

7-10 Inputs generated by the proposed feedback linearization method. . . . . . . . . . . 86

7-11 The actual trajectory followed by the UAV in 3- dimensions . . . . . . . . . . . . . 88

7-12 Vehicle’s trajectory in X,Y and Z . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

7-13 Vehicle’s trajectory in X,Y and Z . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

xii
7-14 The actual orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

7-15 The actual trajectory followed by the UAV in 3- dimensions . . . . . . . . . . . . . 91

7-16 Vehicle’s trajectory in X,Y and Z . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

7-17 Vehicle’s trajectory in X,Y and Z . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

7-18 The actual orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

7-19 Angular velocity around Z axis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

7-20 Snapshot of the hovering with the tilted angle . . . . . . . . . . . . . . . . . . . . 94

7-21 Vehicle’s trajectory in 3-dimension . . . . . . . . . . . . . . . . . . . . . . . . . 95

7-22 The actual orientation of the vehicle along the three directions . . . . . . . . . . . 95

7-23 Vehicle’s trajectory in 3-dimension . . . . . . . . . . . . . . . . . . . . . . . . . 96

7-24 The actual orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

7-25 Vehicle’s position versus time graph along the 3 directions . . . . . . . . . . . . . 97

7-26 Vehicle’s trajectory in 3-dimension . . . . . . . . . . . . . . . . . . . . . . . . . 98

7-27 The actual orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

7-28 Vehicle’s position versus time graph along the 3 directions . . . . . . . . . . . . . 99

xiii
List of Symbols

ACAH . . . . . . . . . . . . . . . . . . . . Attitude Command Attitude Hold

ESC . . . . . . . . . . . . . . . . . . . . . . Electronic Speed Controller
FEM . . . . . . . . . . . . . . . . . . . . . . Finite Element Method
FPGA . . . . . . . . . . . . . . . . . . . . Field Programmable Gate Array
IMU . . . . . . . . . . . . . . . . . . . . . . Inertial Measurement Unit
MV . . . . . . . . . . . . . . . . . . . . . . . Measured values
PWM . . . . . . . . . . . . . . . . . . . . . Pulse Width Modulation
RC . . . . . . . . . . . . . . . . . . . . . . . Remote Control
rpm . . . . . . . . . . . . . . . . . . . . . . . Revolutions per minute
Rx . . . . . . . . . . . . . . . . . . . . . . . . Receive
Tx . . . . . . . . . . . . . . . . . . . . . . . . Transmitt
SP . . . . . . . . . . . . . . . . . . . . . . . . Setpoints for the controller
UART . . . . . . . . . . . . . . . . . . . . Universal Asynchronous Receiver/Transmitter
GW . . . . . . . . . . . . . . . . . . . . . . . Gross Weight
lipo . . . . . . . . . . . . . . . . . . . . . . . lithium polymer
3D . . . . . . . . . . . . . . . . . . . . . . . . Three dimensional
QTW . . . . . . . . . . . . . . . . . . . . . Quad Tilt Wing
QTR . . . . . . . . . . . . . . . . . . . . . . Quad Tilt Rotor
UAV . . . . . . . . . . . . . . . . . . . . . . Unmanned Air Vehicle
VTOL . . . . . . . . . . . . . . . . . . . . Vertical TakeOff and Landing
V/STOL . . . . . . . . . . . . . . . . . . Vertical/Short TakeOff and Landing

xiv
Chapter 1

Introduction

This chapter discusses the objectives, approaches, applications and contributions of the

research. The list of publications resulting from the present work are also provided.

1.1 Objectives

The primary objective of this work is to model, design, control, fabricate and experimen-

tally study quadcopter with tilting propellers. The tilting propellers are expected to result in

more stability as well as the ability to follow any arbitrary trajectory in a smoother manner as

compared to the conventional quadcopters and also add the capability to continue the mission

in the case of one propeller failure.

In addition to this advantage, the tilting mechanism turns the conventional quadcopter,

which is an under-actuated vehicle, to an over-actuated robot which allows full control over

a wide range of state-space. In order to incorporate this new mechanism, which makes the

dynamics of the quadcopter highly nonlinear, this research focuses on developing novel control

mechanisms in order to achieve the desired flight requirements as well as make the vehicle

robust to one propeller failure during the flight.

1.2 Approaches

In order to accomplish the objectives set forth for this study, the new equations of motion for

the tilting quadcopter are first derived mathematically. The derived equations is used to design

1
a linear and nonlinear controller. The numerical simulation of the platform is programmed in

both Simulink and MATLAB for solving the highly nonlinear dynamic equations of motion.

The simulation environment is then used to verify the performance of the developed control

mechanisms. Once the simulation results indicate achievement of desired performance after

running with different flight situations, a 3D model of the new platform is designed using

SolidWorks. The designed 3D model is used to make and 3D print different parts and fabricate

the vehicle. Several attempts are made to improve the platform design. Each attempt focused

on factors such as the weight, the robustness and the functionality of the platform. Several set

of real-world flight tests are carried out to evaluate the capability of the platform in different

flight conditions. The actual flight tests have validated the derived equations for the dynamic

model as well as the designed control systems.

1.3 Motivation for this work

Quadcopters are one of the most popular designs for miniature aerial vehicles (MAVs)

[56]due to their vertical take-off and landing capability, simplicity of construction, maneu-

verability, and ability to negotiate tight spaces making it possible for use in cluttered indoor

areas. Due to these capabilities, quadcopters have recently been considered for a variety of

applications both in military and civilian domains [21, 49, 18, 12]. In particular, quadcopter

MAVs have been explored for applications such as surveillance and exploration of disasters

[11, 33, 67] (such as fire, earthquake, and flood), search and rescue operations [17, 73], moni-

toring of hazmat spills [2], and mobile sensor networks [19, 72].

Blimps [91], fixed-wing planes, single rotor helicopters, bird-like prototypes [28], coaxial

dual rotor helicopters [53], quad-rotors [68, 69, 70, 71] and tilting rotor quadcopters [51, 55,

54, 26, 61, 78] are examples of different configurations and propulsion mechanisms that have

been developed to allow 3D movements in aerial platforms. Each of these has advantages and

drawbacks. This dissertation focuses on quadcopters or quad-rotors which consist of four rotors

in total, with two pairs of counter-rotating, fixed-pitch blades located at the four corners of the

aircraft. This kind of design has two main advantages over the comparable vertical takeoff and

2
landing (VTOL) Unmanned Aerial Vehicles (UAVs) such as single rotor helicopters. Firstly,

quad-rotors do not require complicated mechanical linkage for rotor actuation. Quad-rotors uti-

lize four fixed pitch rotors the variations of whose speeds form the basis of the control. It results

in simplified design and maintenance of the quad-rotors. Secondly, the use of four individual

rotors results in their smaller diameters as compared to the similar main rotor of a helicopter.

The smaller the rotors the less is stored kinetic energy associated with each rotor. This dimin-

ishes the risk posed by the rotors if it comes in contact with any external object. Furthermore,

by securing the rotors inside a frame, the protection of rotors during collisions is achieved. It

allows indoor flights in obstacle-dense environments with lower risk of quad-rotor damage, and

higher operator and surrounding safety. These benefits have resulted in safe test flight by inex-

perienced pilots in indoor environments and recovery time in case of collisions. In particular,

vertical, low speed, and stationary flight are well-known characteristics of a quad-rotors. Struc-

turally, quad-rotors can be made in a small size, with a simple mechanics and control. Though,

as a main drawback, the high energy consumption can be mentioned. However, the trade-off

results are very positive. This configuration can be attractive in particular for surveillance, for

imaging dangerous environments, and for outdoor navigation and mapping.

Conventionally, the quad-rotor attitude is controlled by changing the rotational speed of

each motor. The front rotor and back rotor pair rotates in a clockwise direction, while the

right rotor and left rotor pair rotates in a counter-clockwise direction. This configuration is

devised in order to balance the moment created by each of the spinning rotor pairs. There are

basically four maneuvers that can be accomplished by changing the speeds of the four rotors.

By changing the relative speed of the right and left rotors, the roll angle of the quad-rotor is

controlled. Similarly, the pitch angle is controlled by varying the relative speeds of the front and

back rotors, and the yaw angle by varying the speeds of clockwise rotating pair and counter-

clockwise rotating pair. Increasing or decreasing the speeds of all four rotors simultaneously

controls the collective thrust generated by the robot.

One of the basic limitations of the classical quad-rotor design is that by having only 4

independent control inputs, i.e., the 4 propeller spinning velocities, the independent control of

the six-dimensional position and orientation of the quad-rotor is not possible. For instance, a

3
quad-rotor can hover in place only and if only when being horizontal to the ground plane or it

needs to tilt along the desired direction of motion to be able to move. Tilting rotor quadcopter

concept has evolved to solve these basic limitations of a quad-rotor.

Tilt-design makes the dynamics of the quadcopter more complex, and introduces additional

challenges in the control design. However, tilting rotor quadcopter, designed by using addi-

tional four servo motors that allows the rotors to tilt, is an over-actuated system that potentially

can track an arbitrary trajectory over time. It gives the full controllability over the quad-rotor

position and orientation providing possibility of hovering in tilted configuration.

Another application of the tilting platform lies in its ability to recover during a failure situa-

tion. In conventional quadcopters, if one of the propellers fails, due to its inherent dependency

on the symmetry of the platform, the vehicle becomes entirely uncontrollable. However, in the

proposed platform, if one of the motors completely fail, by using the tilt mechanism of one of

the three remaining motors, the unbalanced momentum can be compensated. Moreover, pro-

viding additional actuation would make the quadcopter more robust to disturbances which can

be rejected more effectively because of the enhanced maneuverability of the quadcopter with

tilting design.

There is a lot of interest recently in developing small aerial vehicles that can carry humans

for transportation in an autonomous manner. To have the quadcopter to be operational for such

purposes, it needs to be scaled up to be able to carry more payloads. Due to safety issues of

the human passenger, it should be robust to external disturbances and unpredictable situations.

To reject external disturbances, agility becomes an important issue. The problem with scaled

up quadcopter is the heavy weight. Once the weight scales up, the inertia will increase and

larger moments will be needed to create the same angular acceleration on a lighter vehicle. As

a consequence, longer propellers would be needed. Therefore, a stronger motor is required to

rotate the new propellers which add more weight to the vehicle. This would make the vehicle

to become more sluggish. Although the tilting mechanism can not make a big difference in

throttle, it is expected that the orientation agility can be increased in big size vehicles due to

the tilting mechanism.

4
1.4 Contribution

This dissertation focuses on designing, fabricating, modeling and controlling a quadcopter

with tilting propellers. The contribution of the work lies in following:

• The mathematical representation of the quadcopter dynamics with tilting rotors has been

derived in order to be used in system analysis and control design.

• Appropriate control techniques have been designed for highly nonlinear dynamics of the

quadcopter with tilting propellers.

• Based on the dynamic equation of motion of tilting rotor quadcopter, the dynamic model

of quadcopter with one motor failure has been derived and the control technique has been

designed in order not only to maintain the stability of the vehicle after the failure, but also

to continue flight mission.

• Two different platforms have been fabricated for the quadcopter which were designed in

SolidWorks environments and some parts have been printed by using 3D printer.

• The numerical simulations and experimental results have validated the mathematical rep-

resentation as well as designed control techniques.

1.5 Publications

Journals

• Alireza Nemati and Manish Kumar. Control of Microcoaxial Helicopter Based on a

Reduced-Order Observer, Journal of Aerospace Engineering, 04015074, 2015.

• Mohammadreza Radmanesh, Manish Kumar, Alireza Nemati and Mohammad Sarim.

Dynamic Optimal UAV Trajectory Planning in the National Airspace System via Mixed

Integer Linear Programming, Proceedings of the Institution of Mechanical Engineers, Part

G: Journal of Aerospace Engineering, 0954410015609361, 2015.

5
• Alireza Nemati and Manish Kumar. Dynamic Modeling and Control of a Quadcopter

with Tilting Rotors, submitted to IEEE Transactions Aerospace and Electronic Systems,

2015.

Book Chapters

• Manish Kumar, Mohammad Sarim and Alireza Nemati. Autonomous Navigation and Tar-

get Geolocation in GPS Denied Environment, Multi-Rotor Platform Based UAV Systems.

Wiley Publishing,

• Manish Kumar, Alireza Nemati, Anoop Sathyan and Kelly Cohen. Real-time Video and

FLIR Image Processing for Enhanced Situational Awareness, Multi-Rotor Platform Based

UAV Systems. Wiley Publishing,

Proceedings

• Alireza Nemati and Manish Kumar. Modeling and Control of a Single Axis Tilting Quad-

copter, American Control Conference (ACC), pp. 3077–3082. IEEE, 2014.

• Alireza Nemati and Manish Kumar. Non-Linear Control of Tilting Quadcopter Using

Feedback Linearization Based Motion Control. Dynamic System and Control Conference

(DSCC), pp. V003T48A005-V003T48A005, ASME, 2014.

• Mohammadreza Radmanesh, Alireza Nemati, Mohammad Sarim and Manish Kumar.

Flight Formation of Quad -copters in Presence of Dynamic Obstacles using Mixed Integer

Linear Programming, Dynamic Systems and Control Conference. ASME , 2015.

• Alireza Nemati, et al. Autonomous Navigation of UAV through GPS-Denied Indoor

Environment with Obstacles, American Institute of Aeronautics and

Astronautics, AIAA SciTech, DOI: 10.2514/6.2015-0715, 2015.

• Mohammad Sarim, Alireza Nemati and Manish Kumar. Autonomous Wall-Following

Based Navigation of Unmanned Aerial Vehicles in Indoor Environments. American

Institute of Aeronautics and Astronautics, AIAA SciTech, DOI: 10.2514/

6.2015-0715, 2015.

6
 • Mohammad Sarim, Alireza Nemati, Manish Kumar and Kelly Cohen. Extended Kalman

Filter based Quadrotor State Estimation based on Asynchronous Multisensor Data, Dy-

namic Systems and Control Conference. ASME , 2015.

• Mohammadreza Radmanesh, Manish Kumar, Alireza Nemati and Mohammad Sarim. So-

lution of Traveling Salesman Problem with Hotel Selection in the Framework of MILP-

Tropical Optimization, accepted in American Control Conference (ACC), IEEE, 2016.

• Alireza Nemati, Neal Soni, Mohammad Sarim, and Manish Kumar. Design, fabrication

and control of a tilt rotor quadcopter. In ASME 2016 Dynamic systems and control con-

ference. American Society of Mechanical Engineers, 2016.

• Alireza Nemati, Rumit Kumar, and Manish Kumar. Stabilizing and control of tilting-rotor

quadcopter in case of a propeller failure. In ASME 2016 Dynamic systems and control

conference. American Society of Mechanical Engineers, 2016.

Intellectual property

• Alireza Nemati, Mehdi Hashemi and Manish Kumar,“ World Frame Based Radio Con-

troller(RC) for Multi-copter UAVs” Filed for provisional patent, October 2015

• Alireza Nemati, Manish Kumar and Rumit Kumar. Fault Tolerance Quadcopter . Provi-

sional Patent Has Been Filed by University of Cincinnati, February 2016.

1.6 Organization of Thesis

This dissertation consists of eight chapters. This includes “Introduction” as the first chapter.

Chapter 2 is a literature review that provides a brief history of the conventional as well as tilting

quadcopters. Chapter 3 reports on the dynamic model of the tilting quadcopter and considers

the nonlinearities that add to the equation due to additional control inputs. Chapter 4 presents a

combined linear and nonlinear controller which is used to control desired orientation during the

flight. The dynamic model of the vehicle and the proposed control technique once the failure

occurs is presented in chapter 5.

7
 The hardware design process is described in detail in Chapter 6. Results from numerical

simulation and experimental studies carried out to verify the modeling and control of tilting

rotor quatcopters following a reference trajectory with simultaneous control of both pitch and

roll angles are discussed in Chapter 7. The conclusions and future works are presented in Chap-

ter 8. This chapter summarizes the dissertation, discusses the contributions and also outlines

directions for future works to be pursued.

8
Chapter 2

History of the Quadcopters

2.1 A Brief History of Quadcopters

2.1.1 The Early History of Quadcopters

A Quadcopter or Quadrotor is multi-rotor mechatronic device capable of Vertical Takeoff

and Landing (VTOL) that is lifted or propelled by four independently rotating rotors. The

idea behind Quadcopters was first developed in the early 1900s. There were very few unique

and momentous quadcopter designs developed throughout the 20th century. The earliest ideas

for a quadcopter were designed and test piloted by Louis Breguet, Etienne Oemichen, George

DeBothezat, and D.H. Kaplan. The first successful flight of a quadcopter aerial vehicle was in

1907. This device, named the Gyroplane (Figure (2-1) 1 ), was built by Breguet brothers and

consisted of a 55hp Renault engine and two forward-tilting 2-blade rotors. It was reported to

have multiple successful flights during the summer of 1908. However it’s mobility and range

of flight were very limited.

1
https://en.wikipedia.org/wiki/Breguet-Richet-Gyroplane

9
 Figure 2-1: Gyroplane.

In the 1920s, Etienne Oemichen, was able to construct the first stable VTOL quadcopter

which he named Oemichen II (Figure (2-2) 2 ). It consisted of a single 180hp Gnome engine

powering four rotors, a complex steel-tube framework of cruciform layout, five smaller pro-

pellers mounted horizontally to provide lateral stability, and an additional pair of propellers

that were mounted to the nose of the craft for steering. The last pair of propellers provided

forward thrust. This design made thousands of successful flights during the mid 1920s and

even established a world record of flying one kilometer in seven minutes and forty seconds.

Almost all quadcopters in the 1920s were unable to sustain a controlled flight and had to use

the Ground Effect to sustain flight limiting these designs to stay low to the ground.

Figure 2-2: The Oemichen2

Around the same time, George DeBothezat designed and built the Flying Octopus (Figure
3
(2-3) for the United States Army Air Corps. The 1678kg ”X” shaped structure supported

2
https://en.wikipedia.org/wiki/C389tienne-Oehmichen
3
https://en.wikipedia.org/wiki/Pescara-Model-3-Helicopter
10
four 8.1m diameter six-blade rotors; one on each end of the 9m long arms. At each end of the

lateral arms, two smaller propellers with variable pitch supplied thrust and enabled yaw control.

After working on his design for a little over two years, DeBothezat was able to develop a fairly

capable quadcopter. This design was able to carry a payload of up to 4 people including the

pilot. However, the design was considered to be flawed as it was under-powered, unresponsive

and very fragile. The craft was only capable of reaching an altitude of around 5m rather than

the 100m desired by the army.

Figure 2-3: De Bothezat

Early quadcopters typically contained a single engine positioned in the center of the fuse-

lage that drove the four rotors via belts or shafts. These belts and shafts were heavy and more

importantly broke down often. In addition, the four rotors of the quadcopter were ever so

slightly different from one another, so the quadcopter was not naturally stable during flight.

Running all rotors at the same speed did not produce a stable flight and each rotor had to be

constantly adjusted to sustain a stable flight. In the early 1900s, with the absence of any dig-

ital computers or sensors, flying a quadcopter required a monumental workload for the pilot

making the early quadcopters very inefficient and not practical for transportation. These early

quadcopters designs also included multiple additional rotors located on different locations of

the quadcopter for additional stability, making these designs not true quadcopters. As mate-

rials and engineering practices evolved over the century, numerous improvements were made

by both increasing the power of the motors and reducing the overall weight of the designs.

During the early 1950s, D.H. Kaplan, worked on and test piloted the Convertawings Model A

11
Quadcopter (Figure (2-4) 4 ). Kaplan’s design featured four rotors and had a two motor layout

with the rotors positioned in an H configuration. Kaplan’s machine may be considered the first

true quadcopter as it was capable of sustaining a controlled flight without the use of the ground

effect or any additional propellers. The 2, 200 pounds craft had a much simpler design than pre-

vious quadcopters due to the fact that control was obtained by varying the thrust between the

individual rotors eliminating the need for complex cyclic-pitch-control systems and additional

rotors on the sides of the fuselage. This design first flew in March of 1956 with great success.

The design, in particular its control system, was a precursor of a majority of the current vertical

takeoff aircraft designs that incorporate tandem wings.

Figure 2-4: Convertawings, Model A

2.1.2 History of Tilt Rotor Vertical Takeoff Vehicles

Tilt Rotors combined the properties of a helicopter which included Vertical Take Off and

Landing (VTOL), hovering, and vertical, forward, and lateral flight, with the desirable prop-

erties of a fixed-wing aircraft including long range flight, low power consumption and the

ability to carry heavier payloads. The first design that resembled a modern tilt rotor device was

patented in May of 1930 by George Lehberger (Figure (2-5))5 .

4
https://en.wikipedia.org/wiki/De-Bothezat-helicopter
5
http://history.nasa.gov/monograph17.pdf

12
 Figure 2-5: George Lehbergers 1930 tilting propeller vertical take-off “flying machine”.

Though this design never amounted to a prototype, it was the first step in making a func-

tional VTOL capable tiltrotor vehicle and inspired the design of the Focke-Achgelis FA-269

(Figure (2-6)6 ) trail-rotor convertiplane project in Germany during World War II[43].

6
http://history.nasa.gov/monograph17.pdf

13
 Figure 2-6: Three-view drawing of the Focke-Achgelis FA-269 convertiplane

A prototype of this aircraft was built in 1943 and consisted of two pusher propellers that

tilted below the two wings for takeoff and landing. However, the project was discontinued

after the allies destroyed a full scale mock-up of this design and much of the research during

a bombing in 1944. A few years later, variants of this tilt rotor configuration surfaced again in

the design studies at Bell and McDonnell Douglas.

The Bell XV-3 (Figure (2-7)7 ) was a tiltrotor aircraft designed by Bell in the 1950s[15].

Its first successful flight was in August 1955. It was the first aircraft to successfully transition

between helicopter and fixed wing for normal flight. The XV-3 was powered by a single 450hp

radial engine that propelled the aircraft at a maximum speed of 296 km/h. The craft had a

maximum altitude of 4600 meters. This aircraft was a proof of concept and made over 100

successful transitions before it was severely damaged in a wind tunnel accident and the design

was scraped.

7
http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20040087005.pdf

14
 Figure 2-7: The Bell XV-3, during flight testing.

The data and experience collected during this trial were key to the development of the Bell-

XV15 (Figure (2-8) 8 ) and the V-22 Osprey (Figure (2-9) 9 )[57, 43].

Figure 2-8: XV-15 taking off

8
https://en.wikipedia.org/wiki/Bell-XV-3
9
https://en.wikipedia.org/wiki/Bell-Boeing-V-22-Osprey

15
 Figure 2-9: The V-22 Osprey, during transition flight

Both these designs followed the same principles as the Bell XV-3 and had many successful

flights over their life.

2.1.3 History of Tilt Rotor Quadcopter

In respect to quadcopter tilting machines, there have been two early designs that stand out,

the Curtiss X-19 and the and the Bell X-22. The Curtiss X-19 (Figure (2-10)10 ) built in 1960s

was a passenger plane that consisted of two sets of thin wings each with a 3 bladed rotor that

could rotate 90 degrees [30]. With its massive 2,200 hp engines, it could carry up to 550kg of

cargo or 4 passengers along with the two crew members. Two turboshaft engines were housed

in the rear fuselage and powered the four rotors. The aircrafts first flight was in 1963 and was

capable of flying up to 523 Km and reached a maximum speed of 650 Km/h. Two prototypes

of the X-19 were built but the project was canceled after the first prototype crashed during its

second flight.

10
https://en.wikipedia.org/wiki/Curtiss-Wright-X-19

16
 Figure 2-10: X-19 in hovering flight

The Bell X-22 was built a couple years after the Curtiss X-19 and is considered to be one of

the most versatile and longest lived of the many VTOL aircrafts that have been developed[30].

It is similar to the X-19 in that it has four wings each with their own 3 bladed propeller each able

to rotate 90 degrees but instead of having 2 motors, the Bell X-22 had four 1250 hp motors each

powering their own rotor. The design was able to carry up to six passengers and two pilots and

reached a maximum speed of 507 km/hour with a range of up to 716km. The two prototypes

of the Bell X-22 were used for many years by both NASA and the US Navy for V/STOL and

performed very well. One is still on display at the Niagara Falls Aerospace Museum in New

York (Figure (2-11)11 ). More modules of quadcopters are currently being developed for the US

Army Corps Including the Bell Boeing Quad Tiltrotor (QTR). It is currently under study and

was first designed in 1999. The Bell Boeing Quad Titlrotor is predicted to be able to carry up

to 80 passengers with a cruise speed of 520 km/hour.

Figure 2-11: Bell X-22A

11
https://en.wikipedia.org/wiki/Bell-X-22

17
2.2 Current Quadcopters

Quadcopters are one of the most popular designs for miniature aerial vehicles (MAVs) due

to their vertical take-off and landing capability, simplicity of construction, maneuverability, and

ability to negotiate tight spaces making it possible for their use in cluttered indoor areas. Due to

these capabilities, quadcopters have recently been considered for a variety of applications both

in military and civilian domains. In particular, quadcopter MAVs have been explored for appli-

cations such as surveillance and exploration of disasters (such as fire, earthquake, and flood),

search and rescue operations, monitoring of hazmat spills, and mobile sensor networks[37]

[64][14]. Blimps, fixed-wing planes, single rotor helicopters, bird-like prototypes, coaxial dual

rotor helicopters, quad-rotors, tilting rotor quadcopters are examples of different configurations

and propulsion mechanisms that have been developed to allow 3D movements in aerial plat-

forms [39] [5] [90] [50] . Each of these designs have advantages as well as drawbacks. This

work focuses on quadcopters or quad-rotors which consist of four rotors in total, with two pairs

of counter-rotating, fixed-pitch blades located at the four corners of the aircraft. This kind of

design has two main advantages over the comparable vertical takeoff and landing (VTOL) Un-

manned Aerial Vehicles (UAVs) such as single rotor helicopters. Firstly, quad-rotors do not

require complicated mechanical linkage for rotor actuation. Quad-rotors utilize four fixed pitch

rotors the variations of whose speeds form the basis of the control. It results in simplified design

and maintenance of the quad-rotors. Secondly, the use of four individual rotors results in their

smaller diameters as compared to the similar single rotor of a helicopter. Smaller rotors imply

less stored kinetic energy associated with each rotor. This diminishes the risk posed by the ro-

tors if it comes in contact with any external object. Furthermore, by securing the rotors inside

a frame, the protection of rotors during collisions is achieved. This configuration allows indoor

flights in obstacle-dense environments with lower risk of quad-rotor damage, and higher oper-

ator and surrounding safety. These benefits have resulted in safe test flights by inexperienced

pilots in indoor environments and lesser recovery time in case of collisions [25]. In particular,

stable, vertical, low speed and stationary flights are well-known characteristics of a quad-rotor.

Structurally, quad-rotors can be designed in a small size, with simple mechanics and control.

The quadcopters have been found to be an attractive choice in particular for surveillance, for
18
imaging dangerous environments, and for outdoor navigation and mapping [59] , [22]. The

major drawback, however, is high energy consumption due to the use of four rotors.

Conventionally, the quad-rotor attitude is controlled by changing the rotational speed of

each motor. The front rotor and back rotor pair rotates in a clockwise direction, while the

right rotor and left rotor pair rotates in a counter-clockwise direction. This configuration is

devised in order to balance the moment created by each of the spinning rotor pairs. There are

basically four maneuvers that can be accomplished by changing the speeds of the four rotors.

By changing the relative speed of the right and left rotors, the roll angle of the quad-rotor is

controlled. Similarly, the pitch angle is controlled by varying the relative speeds of the front and

back rotors, and the yaw angle by varying the speeds of clockwise rotating pair and counter-

clockwise rotating pair. Increasing or decreasing the speeds of all four rotors simultaneously

controls the collective thrust generated by the robot [9].

One of the basic limitations of the classical quad-rotor design is that by having only 4

independent control inputs, i.e., the 4 propeller spinning velocities, the independent control of

the six-dimensional position and orientation of the quad-rotor is not possible. For instance, a

quad-rotor can hover in place only and if only while being horizontal to the ground plane or it

needs to tilt along the desired direction of motion to be able to move. Tilting rotor quadcopter

design has been developed to solve these basic limitations of a quad-rotor. In next section, there

are several examples of recent development about tilting concept.

Recently, there has been a renewed interest in quadcopters from hobbyists, universities, and

corporations across the globe. The renewed interest is due to the many significant technological

advances in sensors and micro-controllers over the past decade that have allowed these once

large machines to be miniaturized to fit in the palm of a hand and be autonomously controlled.

A significant number of quadcopters have been introduced for both military and civilian use as

a result of partnerships between companies and universities that have enabled this quadcopter

UAV revolution. Many companies such as AeroQuad, ArduCopter, DJI, and Parrot AR.Drone

have sparked the interest of hobbyist; Coupled with the DIY (Do It Yourself) and open source

movement, Quadcopter UAVs are more popular and progress in the sector is advancing faster

than ever before.

19
2.2.1 Quadcopters with Tilting Mechanism

Quadcopters with tilting propellers are divided into two categories. Quadcopters which

have the capability of flying both as a conventional quadcopter as well as a fixed wing aerial

vehicle by tilting all propellers in the same direction by the same amount of angle and the

one which is capable of making a tilt in any individual propellers independently. In the first

type of tilting quadcopter( Convertible Quad Tilt Rotor (CQTR)), all propellers are changing

their orientation simultaneously with the same amount of angle. However, variety of platform,

mechanism and control methods have been used by different researchers and universities. Since

all propellers tilt with the same angle, the dynamic equation of the aircraft does not change a lot

and the complexity of the equation is not the most critical obstacle that needs to be tackled for

the designers. The ability of transition between vertical take off and hover flight to horizontal

flight is one the most difficulties the needs to be consider for CQTR. With this ability, aircraft

will be able not only to take off and land at any inappropriate area, but also by taking an

advantage of aerodynamic shape of the wing, it will be able of flying horizontally in long

distances. The flight mode that makes the transition between vertical and horizontal flight has

been gaining remarkable interest among researchers. Numerous innovative CQTR platform

have been studied in very last few years. Papachristos et al [62] from university of Patras,

Greece, have focused on hybrid model predictive flight mode conversion control of CQTR.

Their aircraft platform (shown in Figure (2-12)12 )

Figure 2-12: The Quad-TiltRotor concept, University of Patras.

capable of flying both as a quadcopter as well as fix wing aircraft. They have developed

an innovative control scheme based on hybrid systems theory. An approximation of complete

12
http://www.nt.ntnu.no/users/skoge/prost/proceedings/ecc-2013/data/papers/1271.pdf
20
nonlinear dynamics has been derived and used as a model for control during autonomous mid-

flight conversion. Although they have not flown the aircraft, but by using simulation studies its

been shown that their proposed strategy exceeds the functionality of the flight-modes conver-

sion. The standard NACA2411 airfoil has been selected for the design. The wings with total

span of 1 meter are mounted on the tilting mechanism. The wings are capable of rotating 90

degrees angle.

Another research group has focused on design and control of gas-electric hybrid CQTR

with morphing wing in order to extend the hovering flight up to 3 hours or up to 10 hours of

horizontal flight [13]. They have minimized the mechanical morphing wings and aerodynamic

cost for both high speed and low speed flights. A variety of novel features have been used in

their concept Figure (2-13)13 .

Figure 2-13: CAD design of the CQTR, Istanbul Commerce University.

A gas engine-battery hybrid propulsion is used due to capability of carrying heavy payload

and very long flight duration, plus a carbon composite wings not only to reduce the weight

but also for handling both high speed and low speed flight. The V-type structure of the aircraft

allows to extension of the rear rotors from the center of the mass which ensure the full coverage

of the wing area by the rotor for preventing stalls as well as minimum required speed for

horizontal flight. For the transition, two servo motors have been located next to shaft of rear

and front rotors. The vehicle’s take off weight is 20 kg and the wing-span including the larger

winglets is 2.5 meters. Numerical simulation has been used to validate and understand the

flight behavior and performance of the aircraft. By placing four rotors in two axes with almost

the same level, rotor’s thrust will not cover the most, even if like the previous work, the rotors

13
http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7152278

21
of one the axis extend from the center. The researchers in Beihang University, China, [65] have

proposed a prototype concept of CQTR that place the rotors of the front and rear axis of the

horizontal mode in two different levels. In their concept, for change in configuration, the two

front rotors tilt down to -90 degree, while the two rear rotor tilt up to 90 degree. After the

transition is done, all rotors will be facing front with exactly the same angle but in two different
14
levels. Figure (2-14) shows the configuration of the CQTR in both quadcopter and flying

wing mode.

Figure 2-14: The Configuration of the QTR UAV under research a) Quadcopter mode b)Flying

wing mode, Beihang University.

They have focused on the trajectory tracking control for hovering and acceleration flight of

CQTR by using dynamic inversion. Their scenario is based on 4 strategies as follows: take-

off and reach a certain altitude, find an optimal transition trajectory to not only minimize the

transition time but also not to lose altitude, the next scenario is to keep flying in fixed wing mode

and the ability to change the altitude. During the cruise flight, the rotor speed and the forces

allocated between the rotors and the wings maintain the needed thrust to control altitude and

attitude. Attitude Command Attitude Hold (ACAH) method is being used for attitude control

strategy. The last scenario is to lower the speed and keep tilting back the rotors to their original

angle during the takeoff. Although they have built a prototype version of the vehicle with

dimension of 1.8 meter for the wing span and gross weight of 5.2kg, but numerical simulation

is used to evaluate their proposed control system.

Several researches have introduced many interesting prototype of CQTR, but there are few

14
http://comb.buaa.edu.cn/PUBLICATIONS/PATEERS/2014/48.html
22
groups that validate their control systems by experimental results. Hancer et al [23] presented

a prototype CQTR equipped with robust position controller to track desired trajectory under

aerodynamic and external wind disturbances, as shown in Figure (2-15) 1516 .

Figure 2-15: CQTR with integrated actuators in different flight configurations.

Dryden model has been used to model wind effects which is included in dynamic model.

These disturbances are being estimated by using disturbance observer which is commonly used

in motion control systems. Parametric uncertainties and nonlinear term are also added to ex-

ternal disturbances as a total disturbance. Performance of the hovering flight is verified with

the experimental results. The transition from hovering flight to horizontal cruise flight which

is the most critical part of the experiment is not tested in real world, but trajectory tracking

performance is confirmed with numerical simulations.

Mikami and Uchiyama [47] from Nihon University have validated their concept by nu-

merical simulation as well as experimental results. Due to strong nonlinearities of dynamical

behavior, a linearization method without any approximation has been applied to their control

strategy. This proposed control strategy is being used in both translational and rotational con-

trollers. Figure (2-16) 17 shows an overview of developed CQTR

15
http://research.sabanciuniv.edu/15316/1/cdc10.pdf
16
http://people.sabanciuniv.edu/munel/Publications/JournalPapers/Mechatronics-2012.pdf
17
http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7152364

23
 Figure 2-16: Overview of developed CQTR, Nihon University.

Overall weight of the prototype is 0.48 kg with the length and wing span of 0.8 and 0.7

meter, respectively. Although, due to weight limitation and small size of the battery, the en-

durance of the vehicle is not longer than 15 minutes, but the experimental results verified the

validity of flight control strategy. The vehicle is equipped with micro-computer, flow sensor,

radio module, IMU and ultra-sonic sensor. The experimental results show that the transition

from hover flight to horizontal cruise flight has been accomplished.

Even-though above mentioned air-crafts are using a tilting mechanism for the propellers,

but the under actuation problem of the robot still remains, and in addition to that, these ad-

ditional dynamic of convertible quad tilt rotor is not making the big problem for the control

systems. Also by changing the angle of all propellers simultaneously, translational motion can

not be achieved independently without tilting the aircraft. If the aircraft’s arm had the capabil-

ity of independently tilting about their own axis, the dependency of translation and orientation

problem, could be solved and the system would not be called an under actuated robot. There

have been some attempts to carryout this problem by several groups. Ryll et al. [76] from

Max Plank Institute, Germany, proposed a tilting quadcopter with 8 independent control inputs

that allows the aircraft for independent attitude and position control. They have added four

additional servo motors to allow the propellers to tilt about their axes18 .

18
https://www.semanticscholar.org/paper/Modeling-and-control-of-a-quadrotor-UAV-with-Ryll-
BC3BClthoff/3508f3f9497425a4cbc4bc2d2c2d30df3fc6be77

24
 Figure 2-17: CAD model of the quadcopter with tilting propellers, Max Planck Institute.

The linearized compensation control based on the quadcopter’s dynamic has been used

to achieve 6 DOF of motion control. The tracking of an arbitrary trajectory was the main

focus of the group. They have simplified the dynamic model in order to have a suited model

for control design. Since 8 independent inputs are available, their proposed control design is

over actuated. As in many output tracking control techniques, an appropriate way they solved

the problem was to place the output feedback linearization method. Numerical simulation

was the way they applied the controller presented in [76]. The tracking performance of the

controller was validated and the capability of the proposed method to avoid the singularities

was guaranteed. Another paper has been published from the same authors [77] to validate the

proposed strategies. Although the prototype which is called ”omnicopter” has been tested on a

testing gimbals, but their results show the full controllability over 6 DOF body pose in space.

Another group [81] has tested their prototype on the ball joint rig. Ball joint rig is a device

that can be attached to the underneath of the quadcopter and it can let the quadcopter to orient

around and follow the desired orientation without an actual flight19 .

19
https://www.semanticscholar.org/paper/A-Novel-Actuation-Concept-for-a-Multi-Rotor-UAV-Segui-Gasco-
Al-Rihani/28bdc9144a153c132b37382be32e94d83b1abfed/pdf

25
 Figure 2-18: Vehicle prototype on the ball joint rig flight test, Cranfield University.

The main proposal of the prototype was to improve the performance and fault tolerance of

quadcopter vehicles. They have proposed dual axis tilting propellers which enables gyroscope

torque, differential thrusting and thrust vectoring. Not only the mathematical representation of

the model was modeled and verified by the experimental results, but also a control system was

developed based on PD controller and validated through test on a ball joint rig flight test.

Tilt-design makes the dynamics of the quadcopter more complex, and introduces additional

challenges in the control design. However, tilting rotor quadcopter, designed by using addi-

tional four servo motors that allow the rotors to tilt, is an over-actuated system that potentially

can track an arbitrary trajectory over time. It gives the full controllability over the quad-rotor

position and orientation providing possibility of hovering in a tilted configuration. This work

presents a mathematical dynamic modeling of the tilting rotor quadcopter which provides a

description of the dynamical behavior of the quadcopter as a function of the rotational speeds

of each of the four rotors and their respective tilt-angles. The developed mathematical repre-

sentation of the tilting rotor quadcopter can be used to obtain the position and orientation of

the quad-rotor. The same model can further be used to develop a linear and nonlinear control

strategy via which the speeds of the individual motors and the respective tilt-angles can be ma-

nipulated to achieve the desired motion and configuration.

26
Chapter 3

Dynamic Modeling

Unlike traditional quad-rotor models, which have only four rotatory propellers as the vehi-

cle’s inputs, in tilting rotor quadcopters, there are four more servo motors attached to the each

arm that adds one degree of freedom to each of the propellers, resulting in the tilting motion

along their axes. In this chapter, first the dynamic model of a traditional quad-rotor is described,

then, the equations of motion of a tilting rotor quadcopter are presented.

3.1 Traditional Quad-rotor

Figure (3-1) schematically shows the coordinate system and forces acting on a traditional

quad-rotor. In the 3 dimensional space, the world-frame (E) denotes the fixed reference frame

with respect to which all motions can be referred to and the body-frame (B) is a frame attached

to the center of mass of the vehicle. The rotation of each rotor causes an aerodynamic force

or thrust that acts perpendicular to the plane of rotation of the rotor. In addition to the forces,

each rotor produces a moment perpendicular to the plane of propeller rotation. The moment

produced by a propeller on the vehicle is directed opposite to the direction of rotation of the

propeller, and therefore to cancel out rotation along the Z-axis, the moments for rotor 1 and

3 are set in clockwise (−ZB ) direction and for rotor 2 and 4 are set in counter clockwise (ZB )

direction.

Based on NASA Standard Airplane [35], Euler angle transformations are defined by ψ,

θ and φ which respectively represents the heading, attitude and bank angles also referred to

as yaw, pitch and roll angles. Combined transformation matrix from body coordinate to the

27
world coordinate is obtained by three successive rotations. The first rotation is about X axis,

followed by another rotation about Y axis and the last rotation is about Z axis. For each rotation,

transformation matrix can be written as:

 
 
1 0 0 
 
R x = 0 cosφ −sinφ
 
 
 
0 sinφ cosφ 
 
 
 cosθ 0 sinθ 
 
Ry =  0 1 0 
 

−sinθ 0 cosθ
 
 
cosψ −sinψ 0
 
Rz =  sinψ cosψ 0
 
 
 
0 0 1

Resultant transformation matrix will be given by:

 
cψcθ cψsθsφ − sψcφ cψsθcφ + sψsφ
 
 
REB =  sψcθ sψsθsφ + cψcφ sψsθcφ − cψsφ (3.1)
 
 
 −sθ cθsφ cθcφ 

where cψ and sψ denote cos(ψ) and sin(ψ) respectively, and similarly for other angles.

By obtaining vehicle’s vertical forces in the world frame and writing the equations of mo-

tion based on the Newton second law along the X, Y and Z axes, we can write:

X
m ẍ = Fi (sψsφ + cψsθcφ) − C1 ẋ
X
mÿ = Fi (sψsθcφ − cψsφ) − C2 ẏ (3.2)
X
mz̈ = Fi (cθcφ) − mg − C3 ż

where m is the total mass of quad-rotor, g is the acceleration due to gravity, x, y and z are

quadcopter position in world frame coordinate, C1 , C2 and C3 are drag coefficients. Note that

28
the drag forces are negligible at the low speed. Fi ,(i = 1, 2, 3, 4) are forces produced by the

four rotors as given by the following equation:

Fi = K f ω2i (3.3)

where ωi is the angular velocity of ith rotor and K f is a constant. In addition, Euler equations

are written in order to obtain angular accelerations of the vehicle given by:

I x φ̈ = l(F3 − F1 − C10 φ̇)

Iy θ̈ = l(F4 − F2 − C20 θ̇) (3.4)

Iz ψ̈ = M1 − M2 + M3 − M4 − C30 ψ̇

where l is distance of each rotor from the vehicle’s center of mass. I x , Iy and Iz are moment of

inertia along x, y and z directions respectively. C10 , C20 and C30 are rotational drag coefficients.

Mi , (i = 1, 2, 3, 4) are rotors moment produced by angular velocity of rotors and given by:

Mi = Km ω2i (3.5)

where ωi is the angular velocity of ith rotor and Km is a constant.

Figure 3-1: Schematic diagram showing the coordinate systems and forces acting on the quad-

rotor

During a hovering flight, the quad-rotor not only has zero acceleration and velocity but also

29
needs to have zero pitch and roll angles, i.e. r = r0 , θ = φ = 0, ψ = ψ0 , ṙ = 0, θ̇ = φ̇ = ψ̇ = 0.

At this nominal hover state, the produced force form each propellers must satisfy:

1
Fi = (mg) (3.6)
4

and hence motor speeds are given by:

r
mg
ωi = ωh = (3.7)
4k f

3.2 Tilting Rotor Quadcopters

For a tilting rotor quadcopter, four other variables are added representing the angles of the

quad-rotor arms. Adjustment of these angles results into improved vehicle maneuverability and

capability for hovering at a tilted angle.

To illustrate the motion of the tilting rotors quadcopter, a schematic diagram showing the

forces/moments acting and coordinate frames used in the modeling is provided in Figure (3-2).

As it can be seen from this figure, the propellers are free to tilt along their axes. The planes

shown with dashed lines are the original planes of rotation with zero tilt angles for the respective

propellers. Similarly, the planes shown with the rigid lines are the tilted planes of rotation for

the respective propellers. θi , (i = 1, 2, 3, 4) is the tilted angle of the corresponding propellers. It

may be noted that the forces generated by the propellers are perpendicular to these respective

planes of rotations.

30
 Figure 3-2: Coordinate Frames and Free body diagram of Tilting Quadcopter

The equation governing the acceleration of the center of mass is:

       
       
 ẍ  0   F1 sθ1 − F3 sθ3  C1 ẋ
       
m ÿ =  0  − REB  − C2 ẏ
       
     F 2 sθ1 − F 4 sθ3 

  
  F1 cθ1 + F2 cθ2 + F3 cθ3 + F4 cθ4  C3 ż 
     
z̈ −mg

Using the rotational matrix in (1), equations of motion in world-frame can be rewritten as:

m ẍ = F1 sθ1 cψcθ − F3 sθ3 cψcθ − F4 sθ4 cψsθsφ

+ F2 sθ2 cψsθsφ + F4 sθ4 sψcφ − F2 sθ2 sψcφ

+ F1 cθ1 cψsθcφ + F2 cθ2 cψsθcφ

+ F3 cθ3 cψsθcφ + F4 cθ4 cψsθcφ + F1 cθ1 sψsφ

+ F2 cθ2 sψsφ + F3 cθ3 sψsφ + F4 cθ4 sψsφ − C1 ẋ

mÿ = F1 sθ1 sψcθ − F3 sθ3 sψcθ − F4 sθ4 sψsθsφ

+ F2 sθ2 sψsθsφ − F4 sθ4 cψcφ + F2 sθ2 cψcφ

+ F1 cθ1 sψsθcφ + F2 cθ2 sψsθcφ

+ F3 cθ3 sψsθcφ + F4 cθ4 sψsθcφ − F1 cθ1 cψsφ

− F2 cθ2 cψsφ − F3 cθ3 cψsφ − F4 cθ4 cψsφ − C2 ẏ

31
 mz̈ = −F1 sθ1 sθ + F3 sθ3 sθ − F4 sθ4 cθsφ

+ F2 sθ2 cθsφ + F1 cθ1 cθcφ + F2 cθ2 cθcφ

+ F3 cθ3 cθcφ + F4 cθ4 cθcφ − mg − C3 ż (3.8)

Similarly, the angular accelerations are determined by Euler equations:

I x φ̈ = l(F3 cθ3 − F1 cθ1 − C10 φ̇)

+ (M1 sθ1 − M3 sθ3 ) + (M2 0 + M4 0 )

Iy θ̈ = l(F4 cθ4 − F2 cθ2 − C20 θ̇)

+ (M4 sθ4 − M2 sθ2 ) + (M1 0 + M3 0 )

Iz ψ̈ = l(F1 sθ1 + F2 sθ2 + F3 sθ3 + F4 sθ4 − C30 ψ̇)

+ (M1 cθ1 − M2 cθ2 + M3 cθ3 − M4 cθ4 ) (3.9)

where M 0i , (i = 1, 2, 3, 4) are the tilting moments created by the four servo motors attached

to the end of each arm to enable their tilting motion. It may be noted that these moments are

negligible because the moments produced by the servo motors are used to tilt the arms which are

connected to the main body via mechanical bearings. Neglecting bearing friction, the moments

transmitted to the main body of the quadrotor are negligible. Based on the dynamic model

presented above, we propose the following two Theorems. Note that without lost of generality,

yaw angle is assumed to be zero in following theorems.

Theorem 1: Considering the dynamics of the tilting rotor quadcopter given by Equations

(3.8) and (3.9), and assuming the relationship between the tilting angles of the four rotors

θ1 = −θ3 and θ2 = −θ4 and all rotors having equal rotational speeds, the quadcopter, at an

equilibrium hovering state, achieves a roll angle φ given by φ = θ1 /2 when the pitch angle is

zero, and a pitch angle θ given by θ = θ2 /2 when the roll angle is zero.
Proof: In tilt-hovering, the arm angles of the first and third propellers are tilted by θ1 and
θ3 = −θ1 , respectively. This produces a roll angle φ of the vehicle, and, the equations for linear

32
motion of the quadcopter is given by:
     
 F1 s(θ1 − φ) + F3 s(−θ3 − φ) − F2 sφ − F4 sφ  0 
     
m ẍ
     
mÿ =  0 +
 
  0 

(3.10)
     
    
 mz̈  F c(θ − φ) + F cφ + F c(−θ − φ) + F cφ −mg
1 1 2 3 3 4

For hovering, the accelerations ẍ, ÿ, and z̈ should all be equal to zero. Using the equation

corresponding to the acceleration in X direction, and noting that F1 = F2 = F3 = F4 since

rotational speeds of all rotors are the same, the angle φ can be obtained as:

θ1
φ= (3.11)
2

Similarly to the equation (3.10), if the second and fourth arms are tilted, the equations of motion
can be written as:
     
     
m ẍ  0   0 
  
 
   
mÿ = −F sθ − F sθ + F s(θ − θ) + F s(−θ − θ) +  0  (3.12)
 
   1 3 3 2 2 4 4  
 


 mz̈   F cθ + F c(θ − θ) + F cθ + F c(−θ − θ)  −mg


1 2 2 3 4 4

Similar to above, the angle θ resulted from tilting of the second and forth arms, is given by:

θ2
θ= (3.13)
2

Theorem 2: Considering the dynamics of the tilting rotor quadcopter given by Equations

(3.8) and (3.9), and assuming the relationship between the tilting angles of the four rotors

θ1 = −θ3 and θ2 = −θ4 , the motor speed needed for vehicle for hovering with a tilt angle is

given by:
v
t mg
u
ωi = ωh = when θ = 0
θ1
4k f c
2
and
v
t mg
u
ωi = ωh = when φ = 0 (3.14)
θ2
4k f c
2

33
Proof: In hovering with roll angle and zero pitch angle, the acceleration along z axis is zero,

z̈ = 0. Therefore, using the third row of Equation (3.10), we get:

θ1 X
cos( ) Fi = mg (3.15)
2

Based on Equation (3.3), and noting that each rotor’s angular speed is the same (i.e., F1 = F2 =

F3 = F4 ), the angular speed is given by:

v
t mg
u
ωi = ωh = (3.16)
θ1
4k f c
2

Similarly, considering hovering with pitch angle and zero roll angle, the Equation (3.12) gives:

θ2 X
cos( ) Fi = mg (3.17)
2

Now similar to above, the angular speeds of the rotors is given by:

v
t mg
u
ωi = ωh = (3.18)
θ2
4k f c
2

34
Chapter 4

Control System

4.1 Linear Controller Design

In this chapter, the control strategy of the tilting rotor quadcopter is presented. The aim of

the control strategy is not only control the position of the vehicle to follow an arbitrary trajec-

tory in 3 dimensions, but also to have control over the orientation of the vehicle in hovering as

well as during trajectory tracking.

4.1.1 Proportional Derivative Control

The controller inputs are four independent speeds of propellers and their rotations about the

axes of quadcopter arms. Referring to Figure (3-2) and the two Theorems, it is assumed that

θ1 = −θ3 and θ2 = −θ4 . It may be noted that these constraints, in fact, make the over-actuated

system into fully actuated system (two inputs to tilt the rotors another four inputs for their

rotational speeds make the total number of independent control inputs to be six). For 6 DoF

quadcopter, this results into complete control over its position and orientation. The dynamic

model of the tilting rotor quadcopter, described in (3.8) and (3.9), is used to design the PD

controllers for orientation adjustment and trajectory tracking. Figure (4-1) showes the block

diagram of the control algorithm for orientation and position control during the flight.

35
 Figure 4-1: The block diagram of position and orientation control algorithm

To start tracking a specific trajectory, first, hovering from the initial starting point is neces-

sary. Then, the orientation of the vehicle to a specific pitch or roll angle is obtained. In [46],

the relationship between the rotational speeds of the motors and the deviation of the orienta-

tions from nominal vectors for hovering and navigation is described in detail for conventional

quadcopter. Similar to that approach for the tilting rotor quadcopter, the rotational speeds are

observed as:     
ω1  1 0 −1 1  ωh + ∆ω f 
 des     
     
ωdes  1 1 0 −1  ∆ω 
 2   φ
 =  (4.1)
  
   
ωdes  1 0 1 1   ∆ωθ 
 3     
     
ωdes  1 −1 0 −1  ∆ωψ 
4

where ωdes
i , (i = 1, 2, 3, 4) are the desired angular velocities of the respective rotors. The

hovering speed, ωh , is calculated from Theorem II. The proportional- derivative laws are used

to control ∆ωφ , ∆ωθ , ∆ωψ and ∆ω f which are deviations that result into forces/moments causing

36
roll, pitch, yaw, and a net force along the zB axis, respectively, which are calculated as:

∆ωφ = k p, φ (φdes − φ) + kd, φ (pdes − p)

∆ωθ = k p, θ (θdes − θ) + kd, θ (qdes − q)

∆ωψ = k p, ψ (ψdes − ψ) + kd, ψ (tdes − t) (4.2)

where p, q and t are the component of angular velocities of the vehicle in the body frame. The

relationship between these components and derivatives of the roll, pitch and yaw angles are

provided below [34].

   
 φ̇ 
   
 p
   
 θ̇  = T  q 
   
   
ψ̇
   
t

 
 
1 tanθ.sinφ tanθ.cosφ
 
T = 0
 
 cosφ −sinφ 
 
0 secθ.sinφ secθ.cosφ

The relationship between the tilt angles of individual rotors, given by θides , i = 1, 2..4, and

the reference pitch and roll angles is given by :

    
θ1  1
 des     des 
0 1 0 2φh 
     
  des 
θ2  0
 des  
1 0 1  2θh 
  =     (4.3)
θ3  1 0  ∆φh 
 des     
0 1
     
θdes  0 1 0 1 ∆θ 
  
4 h

where φdes
h and θh are reference roll and pitch angles and ∆φh and ∆θh are orientation devia-
des

tions. Figure (4-2) shows the orientation of the vehicle with respect to the tilted propellers. A

proportional-derivative controller is used to control the orientation deviation using the reference

37
orientation values as:

des
h − φ) + kd, φh (φh
∆φh = k p, φh (φdes ˙ − p)
des
∆θh = k p, θh (θhdes − θ) + kd, θh (θ˙h − q) (4.4)

Figure 4-2: Hovering with tilted arms

The mathematical model of DC servo motor is obtained by the following first order transfer

function that relates the motor angular velocity (rad/s) to input voltage (V) as:

Ω(s) K
= (4.5)
V(s) τs + 1

where τ represents the time constant of the system, and K represents the steady state gain

value. The angular position of the servo motor can be obtained by integrating the motor angular

velocity. The transfer function relating the angular position (rad) and input voltage (V) can be

obtained as:
Θ(s) K K
= = 2 (4.6)
V(s) s(τs + 1) τs + s

The above equation represents a second order transfer function. So, this system is identical

to a second order actuation system. Such systems exhibit a transient response when they are

subjected to external inputs or environmental disturbances. It should be noted that, the transient

response characteristics are one of the most important factors in system design. In general,

38
transfer function of a 2nd order system with input, u(t) and output, y(t) can be expressed as:

y(s) kω2
= 2 (4.7)
u(s) s + 2ζω + ω2

The TGY-210DMH servo motor used in tilt rotor quadrotor system is similar to an actuator

system with damping ratio (ζ) = 0.7, and it has an angular speed of 8 rad/s when operated at

6V and 6.98 rad/s when operating at 4.8V . The natural frequency for mathematical model is

considered to be as 16 rad/s by considering a factor of safety equal to 2, the DC gain has been

considered as unity.

In order to have the quad-rotor track a desired trajectory ri,T , the command acceleration,

r̈ides is calculated from proportional-derivative controller based on position error, as [46]:

(r̈i,T − r̈ides ) + kd,i (ṙi,T − ṙi ) + k p,i (ri,T − ri ) = 0 (4.8)

where ri and ri,T (i = 1, 2, 3) are the 3-dimensional position of the quad-rotor and desired

trajectory respectively. It may be noted that ṙi,T = r̈i,T = 0 for hover.

During the flight of a tilting quadcopter, the orientation of the vehicle needs to be set at

specific pitch or roll. This can be obtained by linearizing the equation of motion that correspond

to the nominal hover states. The nominal hover state (φ = φdes

h = θ1 /2, θ = θhdes = θ2 /2, ψ =

ψT , θ̇ = ψ̇ = φ̇ = 0) corresponds to equilibrium hovering configuration with the reference pitch

or roll angles. The change of the pitch or roll angles are supposed to be small during flight. By

linearizing Equation (3.8) about these nominal hovering states, desired pitch and roll angles to

cause the motion can be derived as given by the following equations :

r¨1 des = 2g(Ae

θdes + Be
φdes + C)

r¨2 des = 2g(De

θdes + Ee
φdes + F)
8k f ωh
r¨2 des = ∆ωF (4.9)
mG

39
where

A = −s(2φdes
h )c(ψT )s(θh ) + s(2θh )c(ψT )s(φh )cθh
des des des des

+ c(φdes
h )c(ψT )cθh cφh + c(2θh )c(ψT )c(θh )c(φh )
des des des des des

B = s(2θhdes )c(ψT )s(θhdes )s(φdes

h ) + s(2θh )s(ψT )s(φh )
des des

− c(2φdes des des

h )c(ψT )s(θh )s(φh )

h ) + c(2φh )s(ψT )c(φh )

− c(2θhdes )c(ψT )s(θhdes )s(2φdes des des

+ c(2θhdes )s(ψT )c(φdes

h )

C = s(2φdes
h )c(ψT )c(θh ) + s(2θh )c(ψT )s(θh )s(φh )
des des des des

h ) + c(2φh )c(ψT )s(θh )c(φh )

− s(2θhdes )s(ψT )c(φdes des des des

+ c(2θhdes )c(ψT )s(θhdes )s(φdes

h ) + c(2φh )s(ψT )s(φh )
des des

+ c(2θhdes )s(ψT )s(φdes

h )

D = −(2φdes
h )c(ψT )s(θh ) + s(2θh )s(ψT )
des des

+ c(2θhdes )s(ψT )c(θhdes )s(φdes

h )

+ c(2φdes des des

h )s(ψT )c(θh )s(φh )

E = s(2θhdes )s(ψT )s(θhdes )c(φdes des

h ) − (2φh )s(ψT )

+ (2φdes des des

h )s(ψT )s(θh )c(φh )

+ (2θhdes )s(ψT )s(θhdes )c(φdes des des

h ) − c(2φh )c(ψT )c(φh )

− c(2θhdes )c(ψT )c(φdes

h )

F = s(φdes
h )c(ψT )c(θh ) + s(2θh )s(ψT )s(θh )s(φh )
des des des des

+ s(θhdes )c(ψT )c(φdes

h ) + c(2φh )s(ψT )s(θh )s(φh )
des des des

+ c(2θhdes )s(ψT )s(θhdes )s(φdes des des

h ) − c(φh )c(ψT )c(φh )

− c(θhdes )c(ψT )c(φdes

h )
θ1 θ2
G = c( )c( )
2 2

φdes and e
where e θdes are respectively the desired deviation in roll and pitch angles from the

h and θh respectively) that are needed for position control when

nominal hovering values (φdes des

40
the orientation is set to be given by the nominal hovering values. Equation (5.14) represents a

φdes and e
pair of two coupled linear equations which are to be solved to obtain the e θdes . The final

desired pitch or roll angles are calculated by:

φdes = e
φdes + φdes
h

θdes = e
θdes + θhdes (4.10)

The desired speeds of the individual rotors are calculated by Equation (5.12). Equation

(5.12) is obtained after determination of Equations (5.13) to (4.10).

4.2 Nonlinear Control

The four rotational velocities of the rotors are the inputs of the vehicle, but in order to

simplify the equations of motion which are described in (3.8) and (3.9), new artificial input

variables are defined as the following. It may be noted that we assume that the tilting happens

only along the roll direction.

u1 = (F1 + F2 + F3 + F4 )/m

u2 = l(F3 − F1 )/I x

u3 = l(F4 − F2 )/Iy

u4 = k(F1 cosθ1 − F2 cosθ2 + F3 cosθ3 − F4 cosθ4 )/Iz (4.11)

where k is force/moment scaling factor.

The equations of motion of the vehicle, considering small angle assumption and tilting

41
along only roll direction (hence, θ1 = −θ3 and θ2 = θ4 = 0), can be obtained as:

1
ẍ = sinθ1 cosθ + u1 cosθ1 cosφh cosφsinθ
2
ÿ = −u1 cosθ1 sinφh cosφ − u1 cosθ1 cosφh sinφ
1
z̈ = −mg − u1 sinθ1 sinθ + u1 cosθ1 cosφh cosφcosθ
2
− u1 cosθ1 sinφh sinφcosθ

φ̈ = (cosθ1 + ksinθ1 )u2

θ̈ = u3

ψ̈ = ku4 (4.12)

where φh is the the desired roll angle that the quadcopter is supposed to tilt during the flight.

4.2.1 Feedback Linearization

A brief review of nonlinear control using feedback linearization method [86, 94, 80, 79]

is presented here. Among the two fundamental design techniques for feedback linearization,

i.e., Input-Output linearization and Input-State linearization, we utilize the Input-Output lin-

earization technique in which we differentiate the output of the systems as many as times as

needed so that the input of the system appears in the last derivative[32, 1]. This technique is a

systematic way to linearize globally part of, or all, the dynamics of system [29]. The following

paragraphs explains how the new/synthetic input v is chosen in order to yield the following

transfer function from the synthetic input to the output y [36]:

Y(s) 1
= γ
V(s) s

where γ, the relative degree, is the last derivative of output so that the physical input appears in

the equation. If this order is less than the system order (n), then there will be internal dynamics

in the feedback linearized system. In cases when internal dynamics appears, the stability of

these dynamics should be also be considered. Here we consider a nonlinear system in the

42
following form:

ẋ = f (x) + g(x)u

y = h(x) (4.13)

where x(∈ Rn ) is the state vector, u(∈ Rm ) represents the control inputs, and y(∈ R p ) stands

for the outputs, f and g are smooth vector fields, and h is a smooth scalar function. A smooth

function is defined as an infinitely differentiable function. The control design process is to find

an integer γ and a state feedback control law,

u = α(x) + β(x)v (4.14)

where α and β are smooth functions. This control law exactly linearizes the map between the

transformed input v and the output y and yields a linear system (linear from the synthetic input v

to the output y). The above idea can be implemented by successively differentiating the output

as:

y(γ) = Lγf h(x) + Lg Lγ−1

f h(x)u (4.15)

where y(γ) represents the γth derivative of y, Lkf h(x) is called the Lie derivative of Lk−1
f h(x)

with respect to field f . Here, if Lg Lγ−1

f h(x) is bounded away from zero for all x, the control law

is given by

1
u= (−Lγf h + v) (4.16)
Lg Lγ−1
f h

The functions α(x) and β(x) in (4.14) can be obtained as:

−Lγf h(x)
α(x) =
Lg Lγ−1
f h(x)
1
β(x) = (4.17)
Lg Lγ−1
f h(x)

43
In order to facilitate the design block, assuming

F(x) = Lγf h(x)

1
G(x)−1 = (4.18)
Lg Lγ−1
f h(x)

Eq. (4.16) can then be written as:

u = G−1 (x)(v − F(x)) (4.19)

We see that the above inversion-based control law has the capability to shape the output

response by simply designing the new control v to get the closed-loop linear system which

finally yields the desired output.

y(γ) = v (4.20)

Once linearization has been achieved, any further control objective such as pole placement

may be easily met using the linear controls theory.

For the nonlinear quadcopter system under consideration in this section, in order to make

the system feedback linearizable, one may consider choosing x, y and z as the output variables.

It can easily be seen that u2 and u3 in (4.11), which are the control inputs representing the pitch

and roll angular accelerations of the vehicle, do not appear in the equation of the outputs. By

successively differentiating of equations of motion till the input terms appear, we can generate

the new control input of the system. It can be seen the new input terms appear in the fourth

derivatives of the outputs as obtained from (4.12):

" #
x (4)
= f (x) x + g(x) x1 g(x) x2 g(x) x3 u
" #
y(4) = f (x)y + g(x)y1 g(x)y2 g(x)y3 u
" #
z(4) = f (x)z + g(x)z1 g(x)z2 g(x)z3 u (4.21)

44
where

f (x) x = −u̇1 θ̇sinθ1 sinθ + u˙1 θ̇cosθ1 cosφh cosφcosθ

− u˙1 φ̇cosθ1 cosφh sinφsinθ

− u˙1 φ̇θ̇cosθ1 cosφh sinφcosθ)

1
g(x) x1 = sinθ1 cosθ + cosθ1 cosφh cosφsinθ
2
1
g(x) x2 = − u1 sinθ1 cosθ − u1 cosθ1 cosφh cosφsinθ
2
g(x) x3 = −u1 cosθ1 cosφh cosφsinθ

f (x)y = 2u˙1 φ̇cosθ1 sinφh sinφ − 2u˙1 φ̇cosθ1 cosφh cosφ

g(x)y1 = −cosθ1 sinφh cosφ − cosθ1 sinφh sinφ

g(x)y2 = 0

g(x)y3 = u1 cosθ1 sinφh cosφ + u1 cosθ1 sinφh sinφ

f (x)z = 2 − u˙1 θ̇sinθ1 cosθ − 2u˙1 θ̇cosθ1 cosφh cosφsicθ

− 2u˙1 φ̇cosθ1 cosφh sinφcosθ

+ 2u1 θ̇φ̇cosθ1 cosφh sinφsinθ

+ 2u˙1 θ̇cosθ1 sinφh sinφsinθ

− 2u˙1 φ̇cosθ1 sinφh cosφcosθ

+ 2u1 θ̇φ̇cosθ1 sinφh cosφsinθ

1
g(x)z1 = − sinθ1 sinθ + cosθ1 cosφh cosφcosθ
2
+ cosθ1 sinφh sinφcosθ
1
g(x)z2 = u1 sinθ1 sinθ − u1 cosθ1 cosφh cosφcosθ
2
+ u1 cosθ1 sinφh sinφcosθ

g(x)z3 = −u1 cosθ1 cosφh cosφcosθ + u1 cosθ1 sinφh sinφcosθ

" #T
u = u¨1 u2 u3

45
where u is the control inputs which control the x, y and z. We choose u as:

 −1  
g(x) x1 g(x) x2 g(x) x2  − f (x) x + v1 
   
   
u = g(x)y1 g(x)y2 g(x)y3  . − f (x)y + v2  (4.22)
   
   
 g(x)z1 g(x)z2 g(x)z3   − f (x)z + v3 
   

The output equation is now given by:

   
 (4)   
 x  v1 
   
y  = v2  (4.23)
 (4)   
   
 (4)   
z v3

We set pseudo inputs terms as:

   
   (4) (3) (2) 
v1   xd − k x1 e x − k x2 e x − k x3 e˙x − k x4 e x 
   
v  =  y(4) − k e(3) − k e(2) − k e˙ − k e  (4.24)
 2   d y1 y y2 y y3 y y4 y  
  
v   z(4) − k e(3) − k e(2) − k e˙ − k e 
3 d z1 z z2 z z3 z z4 x

where e x , ey and ez are errors defined as:e x = x − xd , ey = y − yd and ez = z − zd , [k x1 , ..., k x4 ] ,

[ky1 , ..., ky4 ] and [kz1 , ..., kz4 ] are gains, xd , yd and zd are desired outputs. From (4.24), the error

dynamics are given by:

x − k x1 e x − k x2 e x − k x3 e˙x − k x4 e x = 0
e(4) (3) (2)

y − ky1 ey − ky2 ey − ky3 e˙y − ky4 ey = 0

e(4) (3) (2)

z − kz1 ez − kz2 ez − kz3 e˙z − kz4 e x = 0

e(4) (3) (2)
(4.25)

A PD controller is also design to control the yaw motion, and is given by:

u4 = ψ¨d = kψ1 (ψ˙d − ψ̇) + kψ2 (ψd − ψ) (4.26)

where kψ1 and kψ2 are derivative and proportional gains respectively.

46
4.2.2 PD Based Tilting Angle Controller

Tilting rotor quadcopter is designed by using additional four servo motors attached to the

end of each arm that allow the rotors to tilt. This capability turns the vehicle into an over-

actuated system that potentially can track an arbitrary trajectory over time. During the flight,

as the non-linear control based on the proposed feedback linearization method provides the

amount of pitch and roll required to track an arbitrary trajectory, a PD controller is also designed

to allow the vehicle to either fly or hover with desired orientation or tilting. The relationship

between the tilt angles of the individual rotors, given by θides , i = 1, 2..4, and the reference pitch

and roll angles is given by [51] :

    
θ1  1
     h 
0 1 0  2φ 
     
  h 
θ2  0 1 0 1  2θ 
  
  =     (4.27)
θ3  1 0 ∆φh 
  
0 1
  
     
4
θ  0 1 0 1 ∆θ  
 
h

where φh and θh are reference roll and pitch angles and ∆φh and ∆θh are orientation devia-

tions. A proportional-derivative controller is used to control the orientation deviation using the

reference orientation values as:

∆φh = k p, φh (φh − φ) − kd, φh p

∆θh = k p, θh (θh − θ) − kd, θh q (4.28)

where p, q (and t in the Equation below) are the components of angular velocities of the vehicle

in the body frame. The relationship between these components and derivatives of the roll, pitch

and yaw angles are provided in [74].

It may be noted that this PD controller is designed to desirably control both pitch and roll

angles. However, in this section, for design of feedback linearization method, we assumed the

quadcopter to be tilted just in roll direction. Hence, for the simulation studies carried out at

chapter 7, we set the reference pitch angle to be zero for this controller.

47
4.3 Stability Analysis

The inherently unstable tilting quadcopter dynamics described in (3.8) and (3.9) can be

written in state-space form: Ẋ(t) = f (X(t), U(t)) where U(t) and X(t) are input and state vectors.

x1 = x

x2 = ẋ1 = ẋ

x3 = y

x4 = ẋ3 = ẏ

x5 = z

x6 = ẋ5 = ż

x7 = φ

x8 = ẋ7 = φ̇

x9 = θ

x10 = ẋ9 = θ̇

x11 = ψ

x12 = ẋ11 = ψ̇

" #T
X = x ẋ y ẏ z ż φ φ̇ θ θ̇ ψ ψ̇
" #T
U = F1 F2 F3 F4 θ1 θ2

The state space model Ẋ(t) = f (X(t), U(t)) is not only non-linear but also highly compli-

cated due to tight coupling between different terms. In order to reduce the number of com-

plicated derivative terms involved in further dynamics, the small angle assumption has been

applied to differentiation described in (3.8) and (3.9). We have linearzed the system about an

operating hovering point while tilting along pitch or roll direction. The operating hovering

point Xe is achieved with the input (Ue ) such that f(Xe (t),Ue (t))=0. The linearized system is

given by:
48
 Ẋ(t) = AX(t) + BU(t)

∂f ∂f
The linearization is carried out via calculating the Jacobian matrices as: A = ∂X
and B = ∂U

calculated at operating point: Xe ,Ue . This yields:

 
 
 0 1 0 0 0 0 0 0 0 0 0 0 
 
 
 0 0 0 0 0 0 K1 0 K2 0 0 0 
 
 
 0 0 0 1 0 0 0 0 0 0 0 0 
 
0 0 0 0 0 0 L1 0 0 0 0 0
 
 
 
0 0 0 0 0 1 0 0 0 0 0 0
 
 
 
 0 0 0 0 0 0 M1 0 M2 0 0 0 
A =   ,
 

 0 0 0 0 0 0 0 1 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 1 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 1 

 
 0 0 0 0 0 0 0 0 0 0 0 0 

49
  
 
 0 0 0 0 0 0 
 
 
 A1 A2 A3 A4 A5 A6 
 
 
 0 0 0 0 0 0 
 
 B1 B2 B3 B4 B5 B6 
 
 

 0 0 0 0 0 0 

 
C 
 1 C2 C3 C4 C5 C6 
B =  
 (4.29)
 0 0 0 0 0 0 
 

D 0 D3 0 D5 0 

 1 
 
 0 0 0 0 0 0 
 
 
 0 E2 0 E4 0 E6 
 
 
 0 0 0 0 0 0 
 

 F1 F2 F3 F4 F5 F6 


where:

1
K1 = (−F4 sinθ4 sinθcosφ + F2 sinθ2 sinθcosφ − F1 cosθ1 sinθsinφ
m
− F2 cosθ2 sinθsinφ − F3 cosθ2 sinθsinφ − F4 cosθ2 sinθsinφ)
1
K2 = (−F1 sinθ1 sinθ − F3 sinθ1 sinθ + F4 sinθ2cosθsinφ
m
+ F2 sinθ2 cosθsinφ + F1 cosθ1 cosθcosφ + F2 cosθ2 cosθcosφ

+ F3 cosθ1 cosθcosφ + F4 cosθ2 cosθcosφ)

1
L1 = (−F4 sinθ2 sinφ − F2 sinθ2 sinφ − F1 cosθ1 cosφ
m
− F2 cosθ2 cosφ − F3 cosθ1 cosφ − F4 cosθ2 cosφ)

M1 = F4 sinθ2 cosθcosφ + F2 sinθ2 cosθcosφ − F1 cosθ1 cosθsinφ

− F2 cosθ2 cosθsinφ − F3 cosθ1 cosθsinφ − F4 cosθ2 cosθsinφ)

M2 = −F1 sinθ1 cosθ − F3 sinθ1 cosθcosφ − F4 sinθ2 sinθcosφ

− F2 sinθ2 sinθsinφ − F1 cosθ1 sinθcosφ − F2 cosθ2 sinθcosφ)

− F3 cosθ1 sinθcosφ − F4 cosθ2 sinθcosφ

50
 A1 = sinθ1 + cosθ1 sinθ, A2 = cosθ2 sinθ, A3 = sinθ1 + cosθ1 sinθ

A4 = cosθ2 sinθ, A5 = F1 cosθ1 + F3 cosθ1 − F1 sinθ1 sonθ + F3 sinθ1 sinθ

A6 = −F2 sinθ2 sinθ − F4 sinθ2 sinθ

B1 = −cosθ1 sinφ, B2 = sinθ2 − cosθ2 sinφ, B3 = −cosθ1 sinφ

B4 = sinθ2 − cosθ2 sinφ, B5 = F1 sinθ1 sinφ + F3 sinθ1 sinφ

B6 = F4 cosθ2 + F2 cosθ2 + F2 sinθ2 sinφ + F4 sinθ2 sinφ

C1 = −sinθ1 sinθ + cosθ1 , C2 = sinθ2 sinφ + cosθ2

C3 = −sinθ1 sinθ + cosθ1 , C4 = sinθ2 sinφ + cosθ2

C5 = −F1 cosθ1 sinθ − F3 cosθ1 sinθ − F1 sinθ1 − F3 sinθ1

C6 = F4 cosθ2 sinφ + F2 cosθ2 sinφ − F2 sinθ2 − F4 sinθ2

D1 = −lcosθ1 + ksinθ1 , D3 = lcosθ1 − ksinθ1

D5 = −lF3 sinθ1 + lF1 sinθ1 + kF1 cosθ1 − kF3 cosθ1

E2 = −lcosθ2 − ksinθ2 , E4 = lcosθ2 + ksinθ2

E6 = −lF4 sinθ2 + lF2 sinθ2 + kF4 cosθ2 − kF2 cosθ2

F1 = lsinθ1 + kcosθ1 , F2 = lsinθ2 − kcosθ2 , F3 = lsinθ1 + kcosθ1

F4 = lsinθ2 − kcosθ2 , F5 = lF1 cosθ1 + lF3 cosθ1 − kF1 sinθ1 − kF3 sinθ1

F6 = lF2 cosθ2 + lF4 cosθ2 − kF2 sinθ2 − kF4 sinθ2

Once the system is linearzied, the controllability of the system in the vicinity of equilibrium

points can be analyzed using tools of linear system theory. We analyzed the system’s contolla-

bility for ranges of values on tiltilng angles along pitch or roll directions. As an example, we

provide results for one particular hovering point when F1 = F2 = F3 = F4 = mg

4
, φdes
h = 0, and

θhdes = 200 .

51
  
 
 0 1 0 0 0 0 0 0 0 0 0 0 
 
 
 0 0 0 0 0 0 0.011 0 21.342 0 0 0 
 
 
 0 0 0 1 0 0 0 0 0 0 0 0 
 
0 0 0 0 0 0 −10.340 0 0 0 0 0
 
 
 
0 0 0 0 0 1 0 0 0 0 0 0
 
 
 
 0 0 0 0 0 0 3.762 0 −0.078 0 0 0 
A =   ,
 

 0 0 0 0 0 0 0 1 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 1 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 1 

 
 0 0 0 0 0 0 0 0 0 0 0 0 

 
 
 1 0 0 0 0 0 0 0 0 0 0 0 
 
 
 0 0 1 0 0 0 0 0 0 0 0 0 
 
0 0 0 0 1 0 0 0 0 0 0 0
 
C = 
 

0 0 0 0 0 0 1 0 0 0 0 0
 
 
 

 0 0 0 0 0 0 0 0 1 0 0 0 

 
 0 0 0 0 0 0 0 0 0 0 1 0 

52
  
 
 0 0 0 0 0 0 
 
 
 0.0314 0.0028 0.0314 0.0028 11.8610 −0.0114
 
 
 0 0 0 0 0 0 
 
 0 0.3421 0 0.3421 0 10.2456 
 
 

 0 0 0 0 0 0 

 

 1.0000 0.9397 1.0000 0.9397 −0.0335 0 
B = 
 
 0 
 0 0 0 0 0 
 
−0.2500 0 0.2500 0 −0.0400 0 

 
 
 0 0 0 0 0 0 
 
 
 0 −0.2281 0 0.2281 0 −0.9028
 
 
 0 0 0 0 0 0 
 
0.0250 0.0667 0.0250 0.0667 0.5000 0.4562 
 

The matrices A and B were used to determine the controllability of the system and was

found to be controllable. This means that a feedback control law U(t) = −K f d X(t) can be

designed to stabilize and control the system via pole placement method where K f d is a 6 × 12

feedback control gain matrix.

53
Chapter 5

Fault Tolerant Flight

5.1 Introduction

In this chapter, stability and control of tilting-rotor quadcopter is presented upon failure

of one propeller during flight. The tilting rotor quadcopter provides advantage in terms of

additional stable configurations. On failure of one propeller, the quadcopter has a tendency of

spinning about the primary axis fixed to the vehicle as an outcome of the asymmetry about the

yaw axis. The tilting-rotor configuration is an over-actuated form of a traditional quadcopter

and it is capable of handling a propeller failure, thus making it a fault tolerant system. In this

chapter, a dynamic model of tilting-rotor quadcopter with one propeller failure is derived and a

controller is designed to achieve hovering and navigation capability. The simulation results of

translational and hovering motion are presented.

Multicopters with six or more propellers are also popular as the vehicle is able to maintain

normal flight if one of the propellers fails [48]. But multicopters are costly as compared to

the quadcopters while applications are the same. VTOL UAVs are finding more and more

applications in civilian domain and this changing scenario demands new rules and regulations in

future[63, 89]. System failures are inevitable during flight of UAVs. Propeller or motor failure

is one of the most common failure in case of quadcopters[6]. Currently, the commercial solution

available to deal with propeller failure is emergency parachute which assists in emergency

landing of quadcopters [75].

The operational scenario of quadcopters requires the design of controllers capable of fault

detection, isolation, and diagnosis [3]. Once the failure occurs, the system must be capable of
54
maintaining the stability of the system and complete the mission without much compromise in

system performance. In passive fault tolerant control system (PFTCS) the control algorithm is

designed to achieve a given objective in healthy or faulty situation without changing its control

law [93], whereas In active fault tolerant control system (AFTCS), to preserve the ability of sys-

tem to achieve the objective, the control law is changed according to fault situation [8, 42]. The

fault diagnosis and identification (FDI) block, also termed as diagnosis unit, consists residual

generator and residual evaluation sub-units. A residual is generated by comparing the process

output and the model output, if the residual differs from zero. The residual evaluation compares

it to a threshold to decide and indicate fault. Based on the diagnosis result the reconfiguration

block has to adapt the controller in such a way that the new controller is able to cope with the

faulty process.

Fault Detection and Isolation (FDI) system for actuator faults for an hexacopter vehicle has

been presented in [20] . A diagnostic Thau observer is applied to the hexacopter nonlinear

model to generate residual signals. In the fault-free case, residuals are close to zero, while in

case of a faulty actuator, the value of residuals and fault is detected. Further, Fault isolation

is realized by exploiting the mathematical model of the hexacopter. By quickly detecting the

fault, the control law can be modified to satisfy the closed-loop requirements of the system and

thus making it an active fault tolerant control.

In [48] the control strategy is presented using periodic solutions for a quadcopter experi-

encing one, two opposite, or three complete rotor failures. The strategy employed is to define

an axis, fixed with respect to the vehicle body, and have the vehicle rotate freely about this axis.

By tilting this axis, and varying the total amount of thrust produced, the vehicles position can

be controlled.

Emergency landing procedure of quadcopter has been presented in [38, 41, 24] by using

PID and Backstepping control approach respectively. The strategy is to switch off the propeller

aligned on the same quadcopter axis of the failed propeller. This action converts the quad-

copter configuration into a birotor aerial vehicle. The UAV becomes free to spin in yaw axis

while controlling the remaining attitudes of the UAV and then emergency landing procedure is

exercised.

55
 The tilting-rotor quadcopter is an over-actuated [51, 52, 82] form of a traditional quadcopter

and it is capable of handling a propeller failure, thus making it a fault tolerant system. A

robust, fault tolerant control law and redundant mechanical design of the quadcopter can ensure

safe handling of the quadcopter even after the propeller failure. In this chapter, the tilt rotor

mechanism and PD control of the quadcopter have been used to stabilize the quadcopter after

the propeller failure and thus control all states of the UAV.

5.2 Fault Detection

Fault Detection and Isolation (FDI) system for motor failure is a very important aspect of

fault tolerant control for quadcopters. Quadcopters belong to the class of very fragile aircraft

and if a motor failure occurs, it leads to highly unstable system dynamics. On failure of one

propeller, the quadcopter has a tendency of spinning about the primary axis fixed to the vehicle

as an outcome of the asymmetry about the yaw axis. The second major asymmetry is created

in the roll or pitch plane depending on the corresponding motor failure. If any one of motor

1 or 3 fails, the asymmetry will be in pitch and yaw plane whereas if motor 2 or 4 fails there

would be a roll and yaw asymmetry. A robust Fault Detection and Isolation (FDI) system

can minimize the reaction time for control system reconfiguration and improve the efficiency

of fault tolerant control significantly. As such, this mechanism plays a key role in FTC. FDI

can be implemented with a current sensor that can be used to monitor the amount of current

supplied to each quadcopter motor. The signal from this sensor can be used to take identify the

fault and take further decision for control system reconfiguration. These current sensors fall in

category of arduino energy monitors and are very easily available.

5.3 Dynamic Modeling

Unlike traditional quadcopter models, which have only four rotary propellers as the vehi-

cle’s inputs, in tilting rotor quadcopters, there are four more servo motors attached to the each

arm that adds one degree of freedom to each of the propellers, resulting in the tilting motions

along their axes. The equation of motion of a tilting rotor quadcopter has been discussed in
56
previous sections. In this section, the equations of motion of a tilting rotor quadcopter with one

propeller failure is presented.

5.3.1 Tilting Rotor Quadcopters with One Propeller Failure

When all the propellers of the tilt-rotor quadcopter are working then it yields a stable config-

uration as a result of symmetry of forces and moments. Assuming that one propeller/motor fails

during hovering flight of quadcopter which is located in the pitch plane. Then, the quadcopter

would possess three working propellers and one failed propeller. Once the failure occurs, the

UAV will experience asymmetry about the yaw axis because of M1 , M3 , M4 moments of work-

ing propellers while M2 = 0. Another asymmetry would occur in pitch plane as F2 = 0 and F4

would still have some magnitude. The equations of motion can be modified by putting F2 and

M2 equal to zero.

Once again by using rotational matrix in (3.1), equations of motion in world-frame can be

written as:

m ẍ = F1 sθ1 cψcθ − F3 sθ3 cψcθ − F4 sθ4 cψsθsφ

+ F4 sθ4 sψcφ + F1 cθ1 cψsθcφ + F3 cθ3 cψsθcφ

+ F4 cθ4 cψsθcφ + F1 cθ1 sψsφ + F3 cθ3 sψsφ

+ F4 cθ4 sψsφ − C1 ẋ

mÿ = F1 sθ1 sψcθ − F3 sθ3 sψcθ − F4 sθ4 sψsθsφ

− F4 sθ4 cψcφ + F1 cθ1 sψsθcφ + F3 cθ3 sψsθcφ

+ F4 cθ4 sψsθcφ − F1 cθ1 cψsφ − F3 cθ3 cψcφ

− F4 cθ4 cψsφ − C2 ẏ

mz̈ = −F1 sθ1 sθ + F3 sθ3 sθ − F4 sθ4 cθsφ

+ F1 cθ1 cθcφ + F3 cθ3 cθcφ + F4 cθ4 cθcφ

− mg − C3 ż (5.1)

It should be noted that F2 terms have vanished from the equations which will result in asym-

57
metry because of one propeller failure.

Similarly, the angular accelerations are determined by Euler equations:

I x φ̈ = l(F3 cθ3 − F1 cθ1 − C10 φ̇) + (M1 sθ1 − M3 sθ3 )

+ M4 0

Iy θ̈ = l(F4 cθ4 − C20 θ̇) + M4 sθ4

+ (M1 0 + M3 0 )

Iz ψ̈ = l(F1 sθ1 + F3 sθ3 + F4 sθ4 − C30 ψ̇)

+ (M1 cθ1 + M3 cθ3 − M4 cθ4 ) (5.2)

where M 0i , (i = 1, 2, 3, 4) are the same tilting moments which are created by the four servo

motors attached to the end of each arm to cause a tilt angle. The absence of F2 , M2 should be

noted in pitch and yaw acceleration equations. The components of rotor moment M1 , M3 , M4

would not produce a symmetrical outcome which represent unstable dynamics of quadcopter

upon propeller failure. The available inputs to stabilize and control this system are angular

speed ω1 , ω3 , ω4 of three working rotors and tilt angle θi , (i = 1, 2, 3, 4) of all rotors.

Theorem-III: Considering the dynamics of tilt-rotor quadcopter upon propeller failure

given by Equations (5.1) and (5.2), the quadcopter can be stabilized in yaw and pitch plane if

fourth rotor is tilted by an angle θ4 such that θ4 = c−1 [ω24 /(ω21 cθ1 + ω23 cθ3 )].

Figure 5-1: Free body diagram of tilt-rotor quadcopter upon propeller failure

Proof: When propeller failure occurs the dynamics of the quadcopter are highly non-linear.
58
Thus, we ignore the drag forces and moments generated because of rotor tilt for simplification.

This assumption simplifies the angular acceleration equations and the equations for pitch and

yaw acceleration are given by:

Iy θ̈ = lF4 cθ4 + M4 sθ4

Iz ψ̈ = lF4 sθ4 + M1 cθ1 + M3 cθ3 − M4 cθ4 (5.3)

If the quadcopter has to be stabilized in pitch and yaw plane, θ̈, ψ̈ should be zero. Thus,

equation (5.3) can be equated to zero to solve for θ4 .

lF4 cθ4 = −M4 sθ4

lF4 = −M4 (sθ4 /cθ4 )

lF4 sθ4 − M4 cθ4 = −M1 cθ1 − M3 cθ3 (5.4)

Substituting the value lF4 in equation (5.4) and rearranging the equation:

−M4 (sθ4 sθ4 /cθ4 ) − M4 cθ4 = −M1 cθ1 − M3 cθ3

−M4 (s2 θ4 + c2 θ4 )/(cθ4 ) = −M1 cθ1 − M3 cθ3 (5.5)

Since, s2 θ4 + c2 θ4 = 1 the above equation reduces to the following form:

cθ4 = M4 /(M1 cθ1 + M3 cθ3 )

θ4 = c−1 [M4 /(M1 cθ1 + M3 cθ3 )] (5.6)

The above expression can be re-written in terms of angular velocity by using equation (3.3) :

θ4 = c−1 [ω24 /(ω21 cθ1 + ω23 cθ3 )] (5.7)

This condition should hold for attaining a stable configuration after one propeller failure in the

tilt-rotor quadcopter. Otherwise, the system can not be stabilized or controlled. In fact, once

59
the system is stabilized minor deviation in angular speeds of propellers and rotor tilt angle can

be utilized to maneuver the quadcopter

Theorem-IV: Once propeller failure occurs, the quadcopter can hold a certain altitude if the

angular speed of remaining propellers is increased by:

0
ω1 = ω1 + ω2 /3
0
ω3 = ω3 + ω2 /3
0
ω4 = ω4 + ω2 /3 (5.8)

which means:

ωnew
h = ωh + ω2 /3 (5.9)

Proof: ω2 represents the angular speed of the second propeller at the instant of failure, this

will result in the loss of altitude but the angular speed of three remaining propellers can be

increased by a factor of ω2 /3 in order to compensate for the loss. On the other hand, an extra

compensation component ω4 /cθ4 should come in the equation of fourth rotor to overcome the

tilt effect of the rotor. Thus, angular speed of fourth rotor will be higher as compared to angular

speed of first and third rotor. ω1 , ω3 , ω4 are the increased angular speeds of the propellers in
0 0 0

order to hold the altitude.

Future, the new angular speeds must satisfy equation (5.7) in order to yield a stable config-

uration. We can conclude this statement in the form of equation (5.8):

0 0 0
θ4 = c−1 [ω42 /(ω12 cθ1 + ω32 cθ3 )] (5.10)

60
5.4 Controller Design

In this section, the control strategy of the tilting rotor quadcopter in a case of motor failure

during the flight is presented. Two PD controllers are used due to compensate the unbalance

moments created by an odd number of propellers, and also to stabilize vehicle’s orientation

and make it functional to continue its mission without crash. The vehicle originally has eight

independent inputs which includes four speed of propellers and four tilted angle of each motor

about its axis. In the case of motor failure, two inputs are automatically out of equations. To

make the vehicle compensate the moments of the vehicle, not only the speed of the remaining

propellers needs to be controlled individually, but also the tilted mechanism needs to be set

in a way that compensates the moment from the breakdown motor. In this work, its assumed

that motor two is the one that stopped working during the flight. It should be noted that the

measurement sensor needs to report the failure immediately. Referring to the theorems, the

tilting angle of motor 1 and motor 3 needs to be set at the same orientation and the tilted angle

of the remaining motors can be immediately set according to Theorem III.

To start compensating the unbalanced moment situation after the failure, first, getting back

to hovering is necessary. Then, the orientation of the vehicle to a specific pitch or roll angle is

obtained. In [46], the relationship between the rotational speeds of the motors and the deviation

of the orientations from nominal vectors for hovering and navigation is described in detail

for conventional quadcopter. Based on the new dynamics equation of the vehicle with three

propellers and one tilted servo motor (motors 1 and 3 are assumed to be level), the following

equations are obtained:

 
    ω0 
  1 
 M x  Lk f ω1 −Lk f ω3
 B   0 0
0 0  0 
    ω 
  3
 My  =  0 Lk f ω4 0  0 
0
(5.11)
 B  
0
    ω 
 M B   0   4 
z 0 0 1  
 θ4 

where M xB , MyB and MzB are torque components separated in body frame. It needs to be men-

tioned that these equations are obtained from linearization of equation (5.2) around its nominal

61
hover states while first and third servo motors are assumed not to be tilted. The rotational speed

on each individual motor and the tilted angle of the rear motor are calculated as:

    
ω1  1 −1 0 0 ωh + ∆ω f 
 des     new 
     
ωdes  1 1 0 0  ∆ω 
 3   φ
 = 

(5.12)
  
  
ωdes  1 0 1 0  ∆ωθ 
 4     

θ4 θ4 + ∆θ4 
 des    
0 0 0 1

where ωdes
i , (i = 1, 3, 4) are the desired angular velocities of the respective rotors. θ4 is the tilted

angle that needs to be hold for motor 4 to attaining a stable configuration and is calculated in

Theorem III. The hovering speed, ωnew

h , is calculated from Theorem IV. The proportional-

derivative laws are used to control ∆ωφ , ∆ωθ , ∆θ4 and ∆ω f which are deviations that result into

forces/moments causing roll, pitch, yaw, and a net force along the zB axis, respectively, which

are calculated as:

∆ωφ = k p, φ (φdes − φ) + kd, φ (pdes − p)

∆ωθ = k p, θ (θdes − θ) + kd, θ (qdes − q)

∆θ4 = k p, ψ (ψdes − ψ) + kd, ψ (tdes − t) (5.13)

where p, q and t are the component of angular velocities of the vehicle in the body frame.

During the flight of a tilting quadcopter, the orientation of the vehicle needs to be set level.

This can be obtained by linearizing the equation of motion that correspond to the nominal

hover states. The nominal hover state (φ = θ = 0, ψ = ψT , θ̇ = ψ̇ = φ̇ = 0) corresponds to

equilibrium hovering configuration with the reference pitch or roll angles. The change of the

pitch or roll angles are supposed to be small during flight. By linearizing Equation (3.8) about

these nominal hovering states, desired pitch and roll angles to cause the motion can be derived

as given by the following equations :

r¨1 des = g(cψθdes + sψφdes ) − sψF4 sθ4

r¨2 des = g(sψθdes − cψφdes ) + cψF4 sθ4 (5.14)

62
where θdes and φdes are the desired pitch and roll to be added to the nominal hover states to

move the vehicle to desired trajectory ri,T , the command acceleration, r̈ides is calculated from

proportional-derivative controller based on position error, as [46]:

(r̈i,T − r̈ides ) + kd,i (ṙi,T − ṙi ) + k p,i (ri,T − ri ) = 0 (5.15)

where ri and ri,T (i = 1, 2, 3) are the 3-dimensional position of the quad-rotor and desired

trajectory respectively. It may be noted that ṙi,T = r̈i,T = 0 for hover.

5.5 Linearization and Stability Analysis

The dynamic model of the vehicle with one motor failure can be described with the differ-

ential equations (5.1) and (5.2). This inherently unstable tilting quadcopter dynamics can be

written in state-space form: Ẋ(t) = f (X(t), U(t)) where U(t) and X(t) are input and state vectors

[10].

x1 = x

x2 = ẋ1 = ẋ

x3 = y

x4 = ẋ3 = ẏ

x5 = z

x6 = ẋ5 = ż

x7 = φ

x8 = ẋ7 = φ̇

x9 = θ

x10 = ẋ9 = θ̇

x11 = ψ

x12 = ẋ11 = ψ̇

63
 " #T
X = x ẋ y ẏ z ż φ φ̇ θ θ̇ ψ ψ̇
" #T
U = F1 F3 F4 θ4

Since the quadcopter will not be used for acrobatic maneuvers after the motor failure, the

highly nonlinear and also complicated dynamics, can be simplified with the small angle as-

sumption [87] to cover hovering and moving around with small deviation in orientation. In

order to reduce the number of complicated derivative terms involved in further dynamics, the

small angle assumption has been applied to differentiation described in (5.1) and (5.2). We have

linearized the system about an operating hovering state (φ = θ = 0, ψ = ψT , θ̇ = ψ̇ = φ̇ = 0).

These operating hovering point Xe is achieved with the input (Ue ) such that

f (Xe (t), Ue (t)) = 0 (5.16)

The linearization of the dynamics will result in A and B matrices,

Ẋ(t) = AX(t) + BU(t)

where:

∂ fi (X0 , U0 )
ai, j =
∂X j
∂ fi (X0 , U0 )
bi, j =
∂U j

The linearization is carried out via calculating the Jacobian matrices [40] yields:

64
  
 
 0 1 0 0 0 0 0 0 0 0 0 0 
 
 
 0 0 0 0 0 0 A2,7 0 A2,9 0 0 0 
 
 
 0 0 0 1 0 0 0 0 0 0 0 0 
 
0 0 0 0 0 0 A4,7 0 0 0 0 0
 
 
 
0 0 0 0 0 1 0 0 0 0 0 0
 
 
 
 0 0 0 0 0 0 A6,7 0 A6,9 0 0 0 
A =   ,
 

 0 0 0 0 0 0 0 1 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 1 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 1 

 
 0 0 0 0 0 0 0 0 0 0 0 0 

 
 
 0 0 0 0 
 
 
 A2,1 A2,2 A2,3 A2,4 
 
 
 0 0 0 0 
 
 A4,1 A4,2 A4,3 A4,4 
 
 
 0 0 0 0 
 
 
 A 
 6,1 A6,2 A6,3 A6,4 
B =   (5.17)
 0 0 0 0 
 
 
 A
 8,1 A 8,2 0 0 

 
 0 0 0 0 
 
 
 0 0 A10,3 A10,4

 
 
 0 0 0 0 
 
 
A12,1 A12,2 A12,3 A12,4 

65
where:

1
A2,7 = (−F4 sinθ4 sinθcosφ − F1 cosθ1 sinθsinφ − F3 cosθ3 sinθsinφ
m
− −F4 cosθ4 sinθsinφ)
1
A2,9 = (−F1 sinθ1 sinθ + F3 sinθ3 sinθ − F4 sinθ4cosθsinφ
m
+ F1 cosθ1 cosθcosφ + F3 cosθ1 cosθcosφ − F4 cosθ4 cosθcosφ)
1
A4,7 = (F4 sinθ4 sinφ − F1 cosθ1 cosφ + F3 cosθ3 cosφ − F4 cosθ4 cosφ)
m
A6,7 = F4 sinθ4 cosθcosφ − F1 cosθ1 cosθsinφ − F3 cosθ3 cosθsinφ − F4 cosθ4 cosθsinφ)

A6,9 = −F1 sinθ1 cosθ + F3 sinθ3 cosθcosφ + F4 sinθ4 sinθcosφ

− F1 cosθ1 sinθcosφ − F3 cosθ1 sinθcosφ − F4 cosθ4 sinθcosφ

A2,1 = sinθ1 + cosθ1 sinθcosφ, A2,2 = −sinθ3 + cosθ3 sinθcosφ

A2,3 = cosθ4 sinθcosφ, A2,4 = F4 sinθ4 sinθcosφ,

A4,1 = −cosθ1 sinφ, A4,2 = −cosθ3 cosφ, A4,3 = −sinθ4 cosφ − cosθ4 sinφ

A4,4 = −F4 cosθ4 cosφ + F4 sinθ4 sinφ

A6,1 = −sinθ1 sinθ + cosθ1 cosθcosφ, A6,2 = sinθ3 sinθ + cosθ3 cosθcosφ

A6,3 = −sinθ4 cosθsinφ + cosθ4 cosθcosφ, A6,4 = −F4 cosθ4 cosθsinφ − F4 sinθ4 cosθcosφ

A8,1 = −lcosθ1 + ksinθ1 , A8,2 = lcosθ1 − ksinθ1

A10,3 = lcosθ4 + ksinθ4

A10,4 = −lF4 sinθ4 + kF4 cosθ4

A12,1 = lsinθ1 + kcosθ1

A12,2 = lsinθ3 + kcosθ3

A12,3 = lsinθ4 − kcosθ4

A12,4 = lF4 cosθ4 + lF4 sinθ4

66
 Once the system is linearized, the controllability of the system in the vicinity of equilibrium

points can be analyzed using tools of linear system theory [92].

 
 
 0 1 0 0 0 0 0 0 0 0 0 0 
 
 
 0 0 0 0 0 0 −4.1241 0 8.249 0 0 0 
 
 
 0 0 0 1 0 0 0 0 0 0 0 0 
 
0 0 0 0 0 0 −2.751 0 0 0 0 0
 
 
 
0 0 0 0 0 1 0 0 0 0 0 0
 
 
 
 0 0 0 0 0 0 4.763 0 0 0 0 0 
A =   ,
 

 0 0 0 0 0 0 0 1 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 1 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 0 

 

 0 0 0 0 0 0 0 0 0 0 0 1 

 
 0 0 0 0 0 0 0 0 0 0 0 0 

 
 
 1 0 0 0 0 0 0 0 0 0 0 0 
 
 
 0 0 1 0 0 0 0 0 0 0 0 0 
 
0 0 0 0 1 0 0 0 0 0 0 0
 
C = 
 

0 0 0 0 0 0 0 0 0 0 0 0
 
 
 

 0 0 0 0 0 0 0 0 0 0 0 0 

 
 0 0 0 0 0 0 0 0 0 0 0 0 

67
  
 
 0 0 0 0 
 
 
 0 0 0 0 
 
 
 0 0 0 0 
 
 0 −1.000 −0.865 −2.751
 
 

 0 0 0 0 

 
 1.000 1.000 0.500 −4.763
B = 
 
 0 
 0 0 0 

 
−0.250 0.250 0 0 
 
 
 0 0 0 0 
 
 
 0 0 0.142 −1.136
 
 
 0 0 0 0 
 
0.020 0.020 0.026 1.878 
 

A feedback control law [58] U(t) = −K f d X(t) can be designed to stabilize and control the

system via pole placement method where K f d is a 4 × 12 feedback control gain matrix as below:

 T
 
 0.177 −0.265 0.326 0.007 
 
 
 1.517 −2.079 2.680 0.034 
 
 
 −0.345 0.320 −0.499 −0.058 
 
−1.183 1.028 −2.059 −0.247
 
 
 
−0.675 0.244 −0.091 −0.082
 
 
 
 −4.799 1.442 −0.324 −0.609 
= 
 
Kfd 

 −54.857 29.104 −23.450 −4.416 

 

 −48.255 17.135 −11.818 −3.983 

 

 22.162 −35.742 50.339 −0.500 

 

 34.199 −26.875 31.562 1.374 

 

 0.785 −0.790 0.870 0.099 

 
 3.298 −3.746 4.039 0.389 

68
Figures (5-2) and (5-3) show how position and orientation of quadcopter with one failed motor

remaining stable.

2
X
Y
1.5 Z

1
meter

0.5

-0.5
0 10 20 30 40 50
time(sec)

Figure 5-2: Vehicle’s Position

0.35
φ
0.3
θ
ψ
0.25

0.2
rad

0.15

0.1

0.05

-0.05
0 10 20 30 40 50
time(sec)

Figure 5-3: Vehicle’s Orientation

69
Chapter 6

Hardware design

6.1 Description of the Prototype

The development of a proposed control techniques for tilting quadcopter requires the devel-

opment of an adequate platform for the preliminary experiments. The primary consideration

for the prototype design was to make the quadcopter small and lightweight so that it was able to

carry an extra component required for the tilting mechanism during the flights. The initial con-

figuration and concept of the tilting quadcoter are presented in Figure (6-1). As the fabricating

process was a senior design project, the cost of the vehicle was kept as low as possible.

The structural design of the tilting quadcopter can be divided into two parts: the central

body where all avionics and components are placed, and the tilting mechanism.

Figure 6-1: The CAD model of the tilting quadcopter

70
6.1.1 Central body

As the most important mechanical components of the tilting mechanism are located at the

center area, this part not only needs to have enough space for all components, but also needs

to maintain its symmetry requirement. The central body or core area needs to have space for

essential component such as:

• Four Electronic Speed Controls (ESC) to control each motor

• Communication Hardware

• A power distribution board

• Autopilot

• Bearings, and

• Servo motors.

There are two prototypes which have been designed and fabricated. The core part in the

first prototype is made using two sheets of flame-retardant Garolite which is connected by

plastic bolts and spacer. The flame-retardant Garolite offers excellent strength, low water ab-

sorption and good electrical insulating quality in both humid and dry conditions with maximum

temperature of 2650 F The material in flame-retardant Garolite is fiberglass-cloth with a flame

retardant resin. The hardness also meets Rockwell M110-M115 which is standard hardness

test. The reason to use two sheets is to have all bearings, servo motors and electronic parts in

between. Four additional polycarbonate round tubes are also attached to the frame in order to

make four arms for the quadcopter. Polycarbonate round tube is a cost effective material with

outstanding mechanical, thermal, chemical, physical, and electrical properties. It is light in

weight, has excellent impact strength, heat resistance to 2500 F and most importantly it comes

in a round shape which provides the tilting mechanism with more degree of freedom to rotate.

6.1.2 Tilting Mechanism

The most important features of the tilting quadcopter is the fact that each arm can ro-

tate about its axis independently. In order to have this independency, four additional inputs
71
are needed to rotate each arm to ensure the tilting mechanism. The servos (Figure (6-2)) are

mounted on top of the lower sheet of the central core. Each arm is directly linked to the servo

motor using gears. The gear is hooked around the tube in the middle by two bearings. All eight

bearings (two for each arm) mount between upper and lower sheets. This unique design allows

the arms to rotate separately with any desired angle without mechanical constraint. Figure (6-3)

shows how servos, tubes, bearings and sheets are connected.

Figure 6-2: HS-5087MH HV Digital Micro Servo.

Figure 6-3: The CAD model of components of the tilting mechanism.

Table (6.1) and (6.2) summarize the specifications and the masses of the vehicle’s compo-

nents of the first prototype.

72
 Table 6.1: Specifications of the Prototype

Dimensions 65cm × 65cm

without propellers

Gross weight 2.4kg

Motor Driver ESC 30Amp

Battery Lithium-polymer cells 4S

Propeller 11×47 APC propeller

Autopilot Pixhawk

Motor Motor 850Kv AC2830-358

Servo HS-5087MH Digital servo

Communications MAVLink, MAVROS and Wifi

Figure 6-4: The real model of the first model of tilting Quadcopter.

After ten test flights, due to weight of the vehicle, the motors were not strong enough to

handle the tests and the experimental data was not acceptable. To reduce the weight and also

make the tilt mechanism more agile, second prototype was designed in Autodesk Inventor and

SolidWorks and fabricated by our group. For the second prototype, the 3D robotics framework

was used as the body and four separate tilt mechanisms were designed and built by using a

73
3D printer. The printed 3D mechanism was mounted at the very end of each arm. The parts

were made in a way that perfectly fit the arms to avoid any slinging. The electronics, power,

propellers, motors and the auto pilot are same as the first prototype.

Table 6.2: Vehicle’s component mass details.

Items Weight (g) Quantity Subtotal (g)

Motors 80 4 320.0

Servos 45 4 180.0

Mechanical Gears 20 4 80.0

Prop adapters 8 4 32

Plastic Bolts 6 16 96

ESC 15 4 60

Battery 650 1 650

bearings 35 8 280

Landing gear 50 1 50

Center plates 85 2 170

Controller 38 1 38

Battery Holder 45 1 45

Arms 35 4 140

Wirings and Connectors 150 1 150

Total 2291

This vehicle also has eight control inputs which are used for rotating the four propellers

and tilting mechanism for the arms. It weights 1kg less than the first prototype. The diameter

also does not exceed 55cm. To tilt the motors around the arms, a tilting mechanism has been

located at the end of each arms. Each tilting mechanism consist of three separate parts: i) the

servo motor holder ii) the plate holder iii) and the motor plate which is mounted on the base

via a tilting mechanism.

There are some similar mechanisms available in the market, but the main reason that makes
74
our design more reliable, is the way the servo is connected to the tilting part. In available

versions of the tilting mechanism, the servo is screwed to the mechanism with the same screw

which holds the tilting parts together. The problem comes when the screw is tightened enough

to attach the servo to the tilting part to avoid any sliding. This also pushes the two separate

mechanisms towards each other and makes it harder to tilt. While more force is needed to be

able to make the pressed part to rotate, an extra force which comes from the servo motor, makes

the cog loose. The looseness between the servo and the tilting part contributes to a delay when

the command is received the motor starts tilting. This delay adds the nonlinear parts to the

control system which is not easy to be taken into the equation.

The currently designed tilting mechanism has an extra screw which is placed between the

tilting mechanism and the servo cog. By having this additional screw, the servo cog can be

fastened to the tilting mechanism as much as needed, while the two tilting mechanisms are

attached to each other with separate screw. With this design, not only there is no friction in

tilting mechanism, the servo would never get loose. Figure (6-5) shows the transparent CAD

model of the tilting mechanism. Figure (6-6) shows how the tilting mechanism is mounted at

the end of quadcopter’s arm and how the servo, tilting mechanism and the motor are connected.

In Figure (6-7), both CAD and actual model are presented.

Figure 6-5: The CAD model of transparent tilting mechanism

75
Figure 6-6: The CAD model of tilting mechanism mounted on the arm with the servo and the

motor

Figure 6-7: The CAD and the actual model of the prototype.

As the servo is directly connected to the tilting mechanism with no gears in between. The

smaller and lighter servo which produces the lower torque compared to the previous design can

also be used. This configuration not only eliminates external gears which were used before,

but also reduces the weight of the servo by half. One of the drawback of this mechanism as

compared to the previous one, is the freedom of tilting around the axis. In previous design

there was not any mechanical limitation for the tilting mechanism. Although the previous

design could rotate around the arms up to 360 degrees, it never was our concern. However, this

76
tilting mechanisms is limited to 60 degrees which meets the criteria of the experiments.

6.2 Drive System

6.2.1 Motors

A multi-rotor flying vehicle is more efficient when it is lighter. One of the important criteria

to choose a suitable motor is to use the lightest possible motor which can provide at least

twice as much thrust to lift the vehicle and has the best response for control system in difficult

flight conditions. Brush-less DC motors afford better efficiency and power density compared to

brushed DC motor [31]. Due to higher power density, controllability, minimum requirements

for maintenance, compact size and light weight [60] they have become increasingly popular

and most commonly used motors in MAV technology.

As a result of these advantages, the 850 KV Brush-less DC (BLDC) motor was selected

for the tilting quadcopter. It is also used by most of the 3D Robotics multi-rotors vehicles. It

delivers a very good level of the thrust compared to even bigger size of counterpart 850 KV

motors.

Figure 6-8: 850 kv brushless DC motors

Most brush-less DC motors have three terminals as shown in Figure (6-8). While these

three terminals are connected to stator, the permanent magnets are placed on the rotor such that

the poles are facing the stator. An Electronic Speed Control (ESC) is used to send the command

to control the rotation of the motor. The power output from ESC by using four cells LiPo and

10 × 47 Propeller are shown in Table (6.3).

77
 Table 6.3: Power output from ESC, 4S LiPo, 10x47 Propeller.

25% 50% 70% 100%

Amp (A) 1.83 5.16 8.9 15.9

Wattage (W) 27.7 79.1 133.4 232

Thrust(gr) 235 537 790 1440

6.2.2 Batteries

Lithium Polymer (LiPo) batteries are currently the preferred power sources for most light-

weight, high-current, high-capacity power storage [7]. They offer high energy-storage/weight

ratio and high discharge rate [88]. These batteries use normal lithium ion chemistry with the

polymer separators to provide high discharge rates. It comes with different cell numbers. Each

cell provide 3.7 Volts with internal resistance of approximately 0.03 Ohm [4].

6.3 Avionics

6.3.1 Control Board

A Control Board needs to be selected that would maintain list of requirements to be able

to have control over position and orientation. As there is no control board for tilt mechanism

in the market, in order to apply the proposed control strategy, it needs not only to be an open

source, but also is required to be able to communicate with an out source computing devices.

Furthermore three axis gyroscope and three axis accelerometers are also need to be included.

Another criteria which was taken into the account to choose the control board was the reason-

able price, the size and the weight. All these reasons led us to use PixHawk autopilot [44].

PixHawk is armed with advanced 32 bit ARM processor, micro SD card for logging, Integrated

backup systems and 14 PWM servo outputs which are ideal for our vehicle to be connected to

additional tilting servo motors [45]. Another specification that made us to use PixHawk was

78
the way it could communicate with Robot Operating System (ROS) which is the main coding

environment to apply the control techniques [66]. ROS is a set of software frameworks for

robot software development. ROS also provides low-level device control and message pass-

ing between processes which allows to share data across multiple and commonly specialized

processes [16].

6.3.2 Communications

MAVLink (Micro Air Vehicle Link) is a very lightweight protocol for communicating with

small unmanned vehicle. It is designed as a header-only message marshalling library [27].

It is mostly used for communication between ground station and small unmanned vehicles.

It can be used to transmit all flight data including altitude, attitude, GPS position, air speed,

battery status, way points etc[84, 85, 83]. The transmitted data has the packet structure. The

payload from the packets are called as a MAVLink message which is identifiable by the ID for

each message. The stream of bytes that can be encoded by the ground station can be sent via

Telemetry or USB serial.

MAVROS is also the package in ROS environment that provides communication driver

with MAVLink communication protocol for various autopilots including PixHawk. Not only

all flight data is available in ROS, but also the new commands which are calculated through

control systems can be sent back to the autopilot in real time.

79
Chapter 7

Numerical Simulations and Experimental

Results

7.1 Numerical Simulations

7.1.1 Simulation set-up

To validate the presented dynamic model and the control method, numerical simulations of

the tilting rotor quadcopter were carried out using the MATLAB. The discretized versions of

the dynamic and the controller equations are solved by the Euler method. Here, we provide

the results from one of the simulation scenarios studied. In this scenario, the vehicle’s initial

position was (0.1, 0.8, 0). The final position was set to (0.8, 0.3, 1.5). In this study the SI unit

system is used. In this scenario, the desired pitch or roll angles were modified during the flight

so that both of these angles were simultaneously controlled. Figure (7-1) shows the reference

pitch and roll angle. It can be seen that for time t = 0 sec to t = 1 sec, the reference pitch and

roll angles are zero. At t = 1 sec, the reference pitch angle increases from 00 and reaches the

value of 180 at t = 4 sec and stays with the same until the end of trajectory. At t = 4 sec, the

reference roll angle is also increased from 00 and reaches the value of 120 at t = 7 sec.

80
 20

Pitch (degree)
15

10

0
0 1 2 3 4 5 6 7 8 9 10
time(sec)
15
Roll (degree)

10

0
0 1 2 3 4 5 6 7 8 9 10
time(sec)

Figure 7-1: The reference (commanded) pitch and roll angle

7.1.2 Tilt-Rotor Quadcopter Simulation results

The procedure to accomplish the flight simulation is to have the vehicle take-off from an

initial point vertically till the desired height, and then steer to the destination point with the

horizontal flight. During flight, the orientation of the vehicle is supposed to change according

to reference inputs without losing the height. The quadcopter trajectory in the three dimensional

space from the initial point to the desired destination is shown in Figure (7-2).

1.5
Z(m)

0.5

0
1

0.5
Y(m) 0.6 0.8
0.2 0.4
0 0
X(m)

Figure 7-2: The actual trajectory followed by the UAV in 3-dimensions

81
 Figure (7-3) shows the actual change in the pitch, roll, and yaw angles of the tilting rotor

quadcopter during the flight. It can be seen that the change in the yaw angle is close to zero

while the actual pitch and roll angles closely follow the reference values.

20
θ(degree)

−20
0 1 2 3 4 5 6 7 8 9 10
time(sec)

20
φ(degree)

−20
0 1 2 3 4 5 6 7 8 9 10
time)sec)
−4
x 10
4
ψ(degree)

0
0 1 2 3 4 5 6 7 8 9 10
time(sec)

Figure 7-3: The actual orientation of the vehicle in 3 directions

40 30

20
θ1 (degree)

θ2 (degree)

20
10

0
0
−10

−20 −20
0 2 4 6 8 10 0 2 4 6 8 10
time(sec) time(sec)

20 20

10
θ3 (degree)

θ4 (degree)

0
0

−10
−20
−20

−40 −30
0 2 4 6 8 10 0 2 4 6 8 10
time(sec) time(sec)

Figure 7-4: The angle of each arm during simulation

Figure (7-4) shows how four servomotors modulate the angle of each arm to follow the

referenced orientation commands during the flight. Figure (7-5) shows how the speed of four

motors changes during the flight to track the trajectory and maintain the height of the vehicle.

Figure (7-6) shows the enlarged view of a portion of the Figure (7-6). It can be seen from Figure
82
(7-6) that the angular velocities of rotors stabilize close to t = 1sec (till when the reference tilts

in pitch and roll are zero). At t = 1sec, the reference pitch angle is commanded to gradually

increase to reach a value of 180 at t = 4sec. Corresponding to this, the individual rotor speeds

can be seen to increase from t = 1sec to t = 4sec. At t = 4sec, the reference roll angle is

commanded to gradually increase to reach a value of 120 at t = 7sec. Consequently, there

is further increase in the rotational speeds of the rotors. This increase in motor speed can be

explained by Theorem 2. Comparing Equation (3.14) (for tilted configuration) to Equation

(3.7) (for non-tilted configuration), the theory predicts the need of more rotor speed in tilted

configuration so that the vertical component of force still balances the weight in the tilted

configuration. The simulation results shown in Figure (7-6) just confirms the theory.

150 150

100 100
ω1(rad/s)

ω2(rad/s)

50 50

0 0
0 2 4 6 8 10 0 2 4 6 8 10
time(sec) time(sec)

150 150

100 100
ω3(rad/s)

ω4(rad/s)

50 50

0 0
0 2 4 6 8 10 0 2 4 6 8 10
time(sec) time(sec)

Figure 7-5: The speed of each rotor

83
 115 115

ω1(rad/s)

ω2(rad/s)
110 110

105 105

100 100
2 4 6 8 10 2 4 6 8 10
time(sec) time(sec)

115 115
ω3(rad/s)

ω4(rad/s)
110 110

105 105

100 100
2 4 6 8 10 2 4 6 8 10
time(sec) time(sec)

Figure 7-6: The enlarged view of the figure 8

7.1.3 Feedback Linearization Numerical Simulation

In order to verify the performance of the proposed feedback linearization based control

method of the tilting-rotor quadcopter, numerous numerical simulations were carried out in

MATLAB/Simulink environment. Here, we present the results from one of the simulations.

The simulation comprises of the UAV taking off from the initial position located at (5.0 , 0.0

,0.0 ) and reaching the desired destination located (4.5, 10.0, 10.0), via passing the waypoint

located at (3.0, 4.0, 5.0). Also, during the flight, the orientation of the vehicle is supposed to

change according to the reference input (in this case, the desired roll angle) without deviating

from the desired trajectory. The reference roll angle is set as follows. During the flight, at

t = 5 sec, the reference roll angle is commanded to gradually increase from 0o to reach a value

of 12o at t = 10 sec. The quadcopter is then supposed to move towards the destination with

the commanded roll angle (refer to the bottom plot of Figure (7-9)). The quadcopter trajectory

obtained from the numerical simulation in the three-dimensional space from the initial point to

the waypoint and then to the final position is shown in Figure (7-7).

84
 10
( 4.5, 10.0, 10.0 )
8

z ( meter)
6

0
10
( 3.0, 4.0, 5.0 )
8

4 ( 5.0, 0.0, 0.0 )

2 6
5.5
5
4.5
y ( meter) 0 3.5
4
3
x ( meter)

Figure 7-7: Quadcopter trajectory in three-dimensional space

Figure (7-8) shows the position and orientation of the vehicle during the flight. Figure (7-9)

shows the comparison between the reference roll and the actual roll angle the the vehicle.

10
10
8
8
z (meter)

y (meter)

6
6

4 4

2 2

0 0
0 5 10 15 20 2 3 4 5 6
time(sec) x (meter)

10 0.1

5 0.05
ψ (degree)
θ (degree)

0 0

−5 −0.05

−10 −0.1
0 5 10 15 20 0 5 10 15 20
time(sec) time(sec)

Figure 7-8: Position and orientation of the quadcopter: altitude vs. time (top left), x-position

vs. y-position (top right), pitch vs. time (bottom left), and yaw vs. time (bottom right).

.
85
 15

φ (degree)
10

0 2 4 6 8 10 12 14 16 18 20
time(sec)

15
φh (degree)

10

0 2 4 6 8 10 12 14 16 18 20
time(sec)

Figure 7-9: The reference roll (bottom) and actual roll angle (top) during the flight.

15 15

10 10
u2 (rad/sec2)
u1 (m/sec2)

5 5

0 0

−5 −5
0 5 10 15 20 0 5 10 15 20
time(sec) time(sec)
−3
x 10
15 15

10
u3 (rad/sec2)

u4 (rad/sec )

10
2

5
5

0
0
−5
0 5 10 15 20 0 5 10 15 20
time(sec) time(sec)

Figure 7-10: Inputs generated by the proposed feedback linearization method.

As we chose x, y, and z as the outputs of the feedback linearization method, and u1 , u2 , u3 ,

and u4 as the normalized total lift forces and control inputs for roll, pitch, and yaw respectively,

Figure (7-10) shows that the zero dynamics of the controller are stable ( as the relative degree
86
is smaller than the order of the system).

7.1.4 Fault Tolerant Numerical Simulation

To validate the presented dynamic model in the case of motor failure and the proposed con-

trol technique, numerical simulations of the tilting rotor quadcopter were carried out using the

MATLAB. The discretized versions of the dynamic and the controller equations are solved by

the Euler method. We have completed two different scenarios in order to evaluate the vehicle’s

response in the case of motor failure in following a trajectory as well as hover flight. In both

scenarios, the altitude maintained its desired value.

7.1.4.1 Hover Flight

The first simulation is the hovering task in one spot with motor failure after stable hover

flight. This scenario shows the performance of the controller and highlights the position control

with motor failure. In the first scenario, mission was started by taking off and hovering in the

fixed altitude and in one spot. The initial take of point is located at (0.2, 0.0, 0.0). Figure

(7-11) shows the 3 dimensional path of the flight. As it can be seen, the vehicle is flying in the

neighborhood of the desired spot. The flight has the error in the range of 0.4 meters to hover

around the spot after one of the motors failed.

87
 15

10
position Z

0
0.6
0.4 0.6
0.2 0.4
0.2
0 0
position Y -0.2 -0.2 position X

Figure 7-11: The actual trajectory followed by the UAV in 3- dimensions

In this scenario the vehicle flew for 200 seconds. Figure (7-12) shows the position of vehicle

in each individual axis. At time t = 20 sec, when motor number 2 stops working, the altitude

of the quadcopter drops and after very short time of adjusting, it maintained its altitude to the

end of the flight. The vehicle also has small amount of movement along X and Y axis.

Figure (7-13) shows how the other motors increase the rotational speed in order to com-

pensate the failed motor force to maintain the altitude. Although the rotational speed of the

motors is increased, but the speed limitation of each motor, will not let the vehicle to maintain

the exact previous altitude.

88
 1

X(m)
0

-1
0 50 100 150 200
time(sec)
0.5
Y(m)

-0.5
0 50 100 150 200
time(sec)
20
Z(m)

10

0
0 50 100 150 200
time(sec)

Figure 7-12: Vehicle’s trajectory in X,Y and Z

ω 1 (rad/s)

ω 2 (rad/s)

100 100

50 50

0 0
0 100 200 0 100 200
time(sec) time(sec)
ω 3 (rad/s)

ω 4 (rad/s)

100 100

50 50

0 0
0 100 200 0 100 200
time(sec) time(sec)

Figure 7-13: Vehicle’s trajectory in X,Y and Z

89
 ×10 -3
1

φ(rad)
0

-1
0 50 100 150 200
time(sec)
0.01
θ(rad)

-0.01
0 50 100 150 200
time(sec)
0.4
ψ(rad)

0.2

0
0 50 100 150 200
time(sec)

Figure 7-14: The actual orientation

Figure (7-14) shows how orientation of the vehicle change after the motor failure. As it can

be seen from the Figure (7-14), the yaw angle changes right after the motor failure and remains

in a constant angle. It can easily be notified that the vehicle is not spinning around Z axis after

the motor failure.

7.1.4.2 Tracking a Trajectory

Here, we provide the results from the second simulation we studied. In this scenario, the

vehicle’s initial position was (0.0, 0.0, 0.0). The final position was set to (30.0, 10.0, 10.0). In

this scenario, It can be seen that for time t = 0 sec to t = 10 sec, the quadcopter flew with all

propellers working and it reached its designated altitude at 10 meters. at t = 10 sec, one the

motors stopped working and the proposed control technique were applied immediately. During

the simulation, it is assumed that right after the motor failure, the control system can take over

and apply the control technique for new faulty system.

The quadcopter trajectory in the three dimensional space from the initial point to the desired

destination is shown in Figure (7-15).

90
 15

10
position Z

0
15
10
40
5 30
20
0 10
position Y 0
position X

Figure 7-15: The actual trajectory followed by the UAV in 3- dimensions

Figure (7-16) shows the position of vehicle in each individual axis. At time t − 10 sec, when

motor number 2 stops working, the altitude of the quadcopter drops by 3 meters and after very

short time of adjusting, it maintained its altitude to the end of the flight.

40
X(m)

20

0
0 10 20 30 40 50 60
time(sec)
20
Y(m)

10

0
0 10 20 30 40 50 60
time(sec)
20
Z(m)

10

0
0 10 20 30 40 50 60
time(sec)

Figure 7-16: Vehicle’s trajectory in X,Y and Z

91
 Figure (7-17) shows how the other motors behaved after motor 2 failure. As it can be seen

from the Figure (7-17), to compensate the lost force from motor 2, all working motors speed

up. Although the rotational speed of the motors is increased, but still the altitude is decreased

to 7 meters due to limitation of the amount of rotational speed each motor can provide.
ω 1 (rad/s)

ω 2 (rad/s)
100 100

50 50

0 0
0 20 40 60 0 20 40 60
time(sec) time(sec)
ω 3 (rad/s)

ω 4 (rad/s)

100 100

50 50

0 0
0 20 40 60 0 20 40 60
time(sec) time(sec)

Figure 7-17: Vehicle’s trajectory in X,Y and Z

Once the quadcopter started to adjust the stability after the motor failure, the unbalance

moments caused most amount of change in yaw angle. As it can be seen, the yaw angle started

to increase right after the motor failure.

92
 0.1

φ(rad)
0

-0.1
0 10 20 30 40 50 60
time(sec)
0.02
θ(rad)

-0.02
0 10 20 30 40 50 60
time(sec)
0.4
ψ(rad)

0.2

0
0 10 20 30 40 50 60
time(sec)

Figure 7-18: The actual orientation

Figure (7-18) shows the orientation of the vehicle before and after motor failure.

1.06
θ 4 (rad)

1.05

1.04
0 10 20 30 40 50 60
time(sec)
×10 -3
10
t(rad/s)

-5
0 10 20 30 40 50 60
time(sec)

Figure 7-19: Angular velocity around Z axis

93
7.2 Preliminary Experimental Results

The tilting rotor quadcopter was tested to find out if it was correctly modeled and controlled.

Extensive simulations were carried out to verify the validity of the dynamic model and control

scheme. This section presents the experimental results obtained with the second prototype. The

first experiment is the hovering task in one spot with tilted orientation. This scenario shows the

performance of the controller and highlights the position control with tilted angle. The second

experiment is intended to demonstrate the performance in tracking a simple trajectory while the

orientation keeps the desired value during the flight. In both scenarios, the altitude maintained

its desired value.

7.2.1 Hovering on the spot

In the first experiment, the scenario was started by taking off and hovering in the fixed

altitude and in one spot while the quadcoter tilts by a desired angle. Figure (7-20) shows the

snapshot of the hovering with the tilted angle.

Figure 7-20: Snapshot of the hovering with the tilted angle

The tilting starts at t = 8s and continues till it reaches the desired angle at t = 12s. The

vehicle hovers at the same spot with 0.4 rad tilt along the pitch direction till t = 17s and starts

to reduce the tilt to the horizontal flight. At t = 25s it starts to tilt in the opposite direction

along the pitch with the same angle and maintains the orientation till t = 33s when it starts to

tilt back to the original attitude again.

94
 2

1.5
Z(m)

0.5

0
3

2 3
1 2
1
0
0
Y(m) -1 -1 X(m)

Figure 7-21: Vehicle’s trajectory in 3-dimension

During the flight, the vehicle not only maintains the position at the same spot, but also keeps

the roll and yaw close to zero. The altitude is also constant at 1.5m during the experiment.

2
Pitch(rad)

-2
0 5 10 15 20 25 30 35 40
time(sec)

2
Roll(rad)

-2
0 5 10 15 20 25 30 35 40
time(sec)

2
Yaw(rad)

-2
0 5 10 15 20 25 30 35 40
time(sec)

Figure 7-22: The actual orientation of the vehicle along the three directions

95
7.2.2 Trajectory

7.2.2.1 Line

Similarly, we carried out another experiment where the UAV performed the following oper-

ation: i) it took off and went vertically up till t = 15s and altitude of 1.75m; ii) then the vehicle

tilted along the roll direction till t = 17s; iii) the vehicle was then commanded to moved hori-

zontally along with Y axes in the tilted position till t = 32s; iv) the vehicle was then commanded

to come back to horizontally aligned orientation (i.e., no tilt) till t = 35s; and finally, v) the

vehicle was commanded to land.

1.5
Z(m)

0.5

0
3
3
2
2
1 1
0 0
Y(m) -1 -1 X(m)

Figure 7-23: Vehicle’s trajectory in 3-dimension

The UAV’s trajectory is shown in Figure (7-23) and the plot of pitch, roll, and yaw angles

versus time are shown in Figure (7-24). It should be noted that the spikes in the plots come

from the noisy data by the Optitrack motion caption system.

96
 2

Pitch(rad)
0

-2
0 5 10 15 20 25 30 35 40
time(sec)

0.5
Roll(rad)

-0.5

0 5 10 15 20 25 30 35 40
time(sec)
2
Yaw(rad)

-2
0 5 10 15 20 25 30 35 40
time(sec)

Figure 7-24: The actual orientation

Figure (7-24) shows that the vehicle maintained the desired roll angle while moving for-

ward. During the flight, the pitch and the yaw angle were supposed to be around zero.

0.8

0.6
X(m)

0.4

0.2
0 5 10 15 20 25 30 35 40
time(sec)
2

1
Y(m)

-1
0 5 10 15 20 25 30 35 40
time(sec)
2
Z(m)

0
0 5 10 15 20 25 30 35 40
time(sec)

Figure 7-25: Vehicle’s position versus time graph along the 3 directions

97
7.2.2.2 Box

Another trajectory were carried out to show that the vehicle can track with any desired

orientation angle in the range of operation. The vehicle was supposed to move in a square

shaped path. The vehicle was commanded to take off and hover in the height of 1.8m in t = 8s

and stay at the same altitude till t = 55s. The vehicle moved along X axis first. During the

flight, the pitch angle was supposed to be maintained by 0.5rad from t = 15s till t = 50s.

1.5
Z(m)

0.5

0
2
-1
-0.5 1
0
0.5 0
1
X(m) Y(m)
1.5 -1

Figure 7-26: Vehicle’s trajectory in 3-dimension

As it can be seen from (7-26), during the flight, the altitude is constant after short transient

take-off period.

98
 1

Pitch(rad)
0.5

-0.5
0 10 20 30 40 50 60
time(sec)

1
Roll(rad)
0.5

-0.5
0 10 20 30 40 50 60
time(sec)

1
Yaw(rad)

0.5

-0.5
0 10 20 30 40 50 60
time(sec)

Figure 7-27: The actual orientation

Figure (7-27) shows the orientation of the vehicle during the flight. The vehicle was com-

manded to fly with a desired pitch angle from t = 17 sec to t = 50 sec.

1
X(m)

-1
0 10 20 30 40 50 60
time(sec)
2

1
Y(m)

-1
0 10 20 30 40 50 60
time(sec)
2
Z(m)

0
0 10 20 30 40 50 60
time(sec)

Figure 7-28: Vehicle’s position versus time graph along the 3 directions

99
 During this period of the time, regardless of direction of the vehicle along X or Y axis,

the pitch was commanded to maintain at 0.5 rad. Figure (7-28) clearly shows the vehicle’s

movement along X ,Y and Z versus time in three different plots.

100
Chapter 8

Conclusions and future work

8.1 Conclusions

In this work, the dynamic modeling and control of a tilting rotor quadcopter was presented.

The tilting rotor design was accomplished by having the tilting capability for the four rotors

achieved via four independent motors used to tilt the rotors along their respective tilting axes.

The relationship between the tilting-rotor angles and the quadcopter orientation was derived

using the dynamic model. However, due to added complexity arising from nonlinear relations

with four additional inputs, and the correlation between these inputs and forces, the Newton-

Euler equation of motions become highly nonlinear. The first two theorems were defined in

order to derive simple relationships between the angular rotations along the tilt directions and

the orientation of the vehicle and rotational speeds of the propellers for hovering. By definition

of these theorems, it was shown that this design makes the quadcopter a fully-controlled system

which can track any arbitrary trajectory.

Hovering with controlled pitch and roll angles, and motion with desired orientation are

some of the features of the novel design of the quadcopter system.

This work, furthermore, suggests a PD based method to control the rotational speeds of the

motors responsible for rotating as well as tilting the rotors in order to follow a desired trajectory.

An Input-Output feedback linearization technique has been applied to a tilting quadcopter

to track a trajectory with a desired orientation during the flight. As the behavior of the tilting

quadcopter, affected by the tilting forces and moments, is highly nonlinear, the linearization

turns out to be a proper control technique to avoid complex behavior of the dynamics. How-
101
ever, in the presence of unmodeled dynamics or any other undefined disturbances, the system

still remains nonlinear after feedback linearization loop. Theoretical results of this control

technique have been supported by simulation results in MATLAB environment.

The dynamic modeling and control of a tilting rotor quadcopter in the case of one motor

failure was also been presented. The relationship between the tilting-rotor angles and the quad-

copter orientation was derived using the dynamic model. It was shown that the quadcopter can

be stabilized if the rotor diagonally opposite to the failed rotor is tilted by an angle that has

been calculated by the Euler equations. In order to maintain an altitude to the closest possible

altitude after the failure, the rotational speed of each individual motor has also been calculated.

The paper presents the dynamic model, and suggests a PD based control in order to avoid crash-

ing of quadcopter and continue the mission. The model and the controller are verified with the

help of numerical simulations.

Two different vehicles with tilting mechanism were designed and fabricated as a prototype

of the concept. The proposed control scheme was implemented on the prototype during labo-

ratory experiments which demonstrated the capability of the the vehicle to hover and navigate

with tilted orientation. Furthermore, such over-actuated systems promise to provide mecha-

nisms to not only overcome wind disturbances more effectively but also provide tolerance to

individual rotor failure.

8.2 Future Works

The work presented here addresses the dynamic modeling and control of a tilting rotor

quadcopter with design and fabrication of two different prototypes. Future work should be

directed towards the following directions:

• Future work could focus on developing modeling and the control techniques when all

rotors can tilt around their axes independently in order to make the flight smoother and

possibly faster in complicated maneuvers.

• Based on our experience with the second prototype, which is lighter and more practically

applicable than the first one, and also by the experience our group had in the other projects
102
 with 3D printed quadcopter frame, a third generation prototype of the quadcopter with

tilting rotors can be a fabricated by a combination of 3D printed parts and the carbon fiber

materials to make the vehicle lighter, firmer and smaller.

• Such over-actuated systems promise to provide mechanisms to overcome wind distur-

bances in outdoor application more effectively. Further work is needed to develop control

methods that can robustly reject such wind disturbances.

• In order to decrease the reaction time of the tilting mechanism, actuation system improve-

ment as well as higher computational processor and better on-board sensors can be inves-

tigated.

• Further studies are required for understanding the time delay between the time of motor

failure and when the new controller takes over the system. Furthermore, this logic can be

implemented in the on board flight software for all the propellers.

• Implement nonlinear control theory to allow the prototype with one motor failure, to track

complex trajectory.

• Experimental work for executing autonomous flight with tilting rotors also forms a future

work. This would require developing auto-pilot from scratch since available auto-pilot

systems are developed for traditional quadcopters that attempt to modify rotor speeds to

make the frame horizontal when they receive IMU feedback that quad is tilted. Some

progress in this respect has been made. The commands which are required to be sent to

the servo motors with respect to the output of designed controller in this work has been

modified in the code in two different auto pilot boards for the tilt rotor quadcopter. For

the experiment related to the motor failure, the KK2.1 flight controller has been used. The

experiments were done by taking off while the quadcopter had three working motors. This

experiment can be performed as a future work in the way that starts with the quadcopter

with all working motors. Subsequently, fault is induced in one of the motors mid-flight

that requires switching to the failure mode control algorithm in an online fashion. This

experiment would ensure that the transition between different control algorithms is smooth

and does not lead to instability.

103
References
[1] K. Ahmadi and E. Salari. Small dim object tracking using a multi objective particle swarm

optimisation technique. IET Image Processing, 9(9):820–826, 2015.

[2] J. Allen and B. Walsh. Enhanced oil spill surveillance, detection and monitoring through

the applied technology of unmanned air systems. In International oil spill conference,

volume 2008, pages 113–120. American Petroleum Institute, 2008.

[3] M. H. Amoozgar, A. Chamseddine, and Y. Zhang. Experimental test of a two-stage

kalman filter for actuator fault detection and diagnosis of an unmanned quadrotor heli-

copter. Journal of Intelligent & Robotic Systems, 70(1-4):107–117, 2013.

[4] D. Andrea. Battery Management Systems for Large Lithium Ion Battery Packs. Artech

house, 2010.

[5] R. W. Beard, D. Kingston, M. Quigley, D. Snyder, R. Christiansen, W. Johnson,

T. McLain, and M. Goodrich. Autonomous vehicle technologies for small fixed-wing

uavs. Journal of Aerospace Computing, Information, and Communication, 2(1):92–108,

2005.

[6] L. Besnard, Y. B. Shtessel, and B. Landrum. Quadrotor vehicle control via sliding mode

controller driven by sliding mode disturbance observer. Journal of the Franklin Institute,

349(2):658–684, 2012.

[7] O. Bilgen, K. Kochersberger, E. C. Diggs, A. J. Kurdila, and D. J. Inman. Morphing

wing micro-air-vehicles via macro-fiber-composite actuators. In Proceedings of the 48th

AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference,

Honolulu, HI, Apr, pages 23–26, 2007.

104
 [8] M. Blanke, M. Staroswiecki, and N. E. Wu. Concepts and methods in fault-tolerant con-

trol. In American Control Conference, 2001. Proceedings of the 2001, volume 4, pages

2606–2620. IEEE, 2001.

[9] S. Bouabdallah, P. Murrieri, and R. Siegwart. Towards autonomous indoor micro vtol.

Autonomous Robots, 18(2):171–183, 2005.

[10] W. L. Brogan. Modern control theory. Pearson Education India, 1974.

[11] D. W. Casbeer, D. B. Kingston, R. W. Beard, and T. W. McLain. Cooperative forest

fire surveillance using a team of small unmanned air vehicles. International Journal of

Systems Science, 37(6):351–360, 2006.

[12] A. Cavoukian. Privacy and drones: Unmanned aerial vehicles. Information and Privacy

Commissioner of Ontario, Canada, 2012.

[13] E. Cetinsoy. Design and control of a gas-electric hybrid quad tilt-rotor uav with morphing

wing. In Unmanned Aircraft Systems (ICUAS), 2015 International Conference on, pages

82–91. IEEE, 2015.

[14] H. Chao, Y. Cao, and Y. Chen. Autopilots for small unmanned aerial vehicles: a survey.

International Journal of Control, Automation and Systems, 8(1):36–44, 2010.

[15] D. A. Conner, B. D. Edwards, W. A. Decker, M. A. Marcolini, and P. D. Klein.

Nasa/army/bell xv-15 tiltrotor low noise terminal area operations flight research program.

Journal of the American Helicopter Society, 47(4):219–232, 2002.

[16] S. Cousins. Exponential growth of ros [ros topics]. IEEE Robotics & Automation Maga-

zine, 1(18):19–20, 2011.

[17] P. Doherty and P. Rudol. A uav search and rescue scenario with human body detection

and geolocalization. In AI 2007: Advances in Artificial Intelligence, pages 1–13. Springer,

2007.

[18] C. P. Ellington. The novel aerodynamics of insect flight: applications to micro-air vehi-

cles. Journal of Experimental Biology, 202(23):3439–3448, 1999.

105
[19] A. T. Erman, L. v. Hoesel, P. Havinga, and J. Wu. Enabling mobility in heterogeneous

wireless sensor networks cooperating with uavs for mission-critical management. Wire-

less Communications, IEEE, 15(6):38–46, 2008.

[20] A. Freddi, S. Longhi, A. Monteriu, and M. Prist. Actuator fault detection and isola-

tion system for an hexacopter. In Mechatronic and Embedded Systems and Applications

(MESA), 2014 IEEE/ASME 10th International Conference on, pages 1–6. IEEE, 2014.

[21] S. Griffiths, J. Saunders, A. Curtis, B. Barber, T. McLain, and R. Beard. Obstacle and

terrain avoidance for miniature aerial vehicles. In Advances in Unmanned Aerial Vehicles,

pages 213–244. Springer, 2007.

[22] S. Grzonka, G. Grisetti, and W. Burgard. A fully autonomous indoor quadrotor. Robotics,

IEEE Transactions on, 28(1):90–100, 2012.

[23] C. Hancer, K. T. Oner, E. Sirimoglu, E. Cetinsoy, and M. Unel. Robust position control

of a tilt-wing quadrotor. In Decision and Control (CDC), 2010 49th IEEE Conference on,

pages 4908–4913. IEEE, 2010.

[24] K. hmadi, A. Y. Javaid, and E. Salari. An efficient compression scheme based on adaptive

thresholding in wavelet domain using particle swarm optimization. Signal Processing:

Image Communication, 32:33–39, 2015.

[25] G. M. Hoffmann, H. Huang, S. L. Waslander, and C. J. Tomlin. Quadrotor helicopter

flight dynamics and control: Theory and experiment. In Proc. of the AIAA Guidance,

Navigation, and Control Conference, volume 2, 2007.

[26] M.-D. Hua, T. Hamel, and C. Samson. Control of vtol vehicles with thrust-tilting aug-

mentation. IFAC Proceedings Volumes, 47(3):2237–2244, 2014.

[27] A. S. Huang, E. Olson, and D. C. Moore. Lcm: Lightweight communications and mar-

shalling. In Intelligent robots and systems (IROS), 2010 IEEE/RSJ international confer-

ence on, pages 4057–4062. IEEE, 2010.

106
[28] V. Hugel, A. Abourachid, H. Gioanni, L. Mederreg, M. Maurice, O. Stasse, P. Bonnin,

and P. Blazevic. The robocoq project: Modelling and design of a bird-like robot equipped

with stabilized vision. In Proceedings of the 2nd International Symposium on Adaptive

Motion of Animals and Machines (AMAM), 2003.

[29] A. Isidori. Nonlinear control systems, volume 1. Springer, 1995.

[30] D. R. Jenkins, T. Landis, and J. Miller. American x-vehicles: An inventory-x-1 to x-50.

2005.

[31] J. G. Kassakian, H.-C. Wolf, J. M. Miller, and C. J. Hurton. Automotive electrical systems

circa 2005. Spectrum, IEEE, 33(8):22–27, 1996.

[32] H. K. Khalil and J. Grizzle. Nonlinear systems, volume 3. Prentice hall New Jersey, 1996.

[33] D. Kingston, R. W. Beard, and R. S. Holt. Decentralized perimeter surveillance using a

team of uavs. IEEE Transactions on Robotics, 24(6):1394–1404, 2008.

[34] A. Ö. Kivrak. Design of control systems for a quadrotor flight vehicle equipped with

inertial sensors. Atilim University, December, 2006.

[35] L.-C. Lai, C.-C. Yang, and C.-J. Wu. Time-optimal control of a hovering quad-rotor

helicopter. Journal of Intelligent and Robotic Systems, 45(2):115–135, 2006.

[36] D. Lee, H. J. Kim, and S. Sastry. Feedback linearization vs. adaptive sliding mode control

for a quadrotor helicopter. International Journal of control, Automation and systems,

7(3):419–428, 2009.

[37] D.-J. Lee, I. Kaminer, V. Dobrokhodov, and K. Jones. Autonomous feature following for

visual surveillance using a small unmanned aerial vehicle with gimbaled camera system.

International Journal of Control, Automation and Systems, 8(5):957–966, 2010.

[38] V. Lippiello, F. Ruggiero, and D. Serra. Emergency landing for a quadrotor in case of a

propeller failure: A pid based approach. In Safety, Security, and Rescue Robotics (SSRR),

2014 IEEE International Symposium on, pages 1–7. IEEE, 2014.

107
[39] Y. Liu, Z. Pan, D. Stirling, and F. Naghdy. Control of autonomous airship. In Robotics

and Biomimetics (ROBIO), 2009 IEEE International Conference on, pages 2457–2462.

IEEE, 2009.

[40] L. Ljung. System identification toolbox. The MathWorks Inc., South Natick, MA, USA,

1988.

[41] R. Luxman and X. Liu. Implementation of back-stepping integral controller for a gesture

driven quad-copter with human detection and auto follow feature. In Computer Science,

Computer Engineering, and Social Media (CSCESM), 2015 Second International Con-

ference on, pages 134–138. IEEE, 2015.

[42] M. Mahmoud, J. Jiang, and Y. Zhang. Active fault tolerant control systems: stochastic

analysis and synthesis, volume 287. Springer Science & Business Media, 2003.

[43] M. D. Maisel, D. J. Giulianetti, and D. C. Dugan. The history of the xv-15 tilt rotor

research aircraft from concept to flight. 2000.

[44] L. Meier, P. Tanskanen, F. Fraundorfer, and M. Pollefeys. Pixhawk: A system for au-

tonomous flight using onboard computer vision. In Robotics and automation (ICRA),

2011 IEEE international conference on, pages 2992–2997. IEEE, 2011.

[45] L. Meier, P. Tanskanen, L. Heng, G. H. Lee, F. Fraundorfer, and M. Pollefeys. Pixhawk:

A micro aerial vehicle design for autonomous flight using onboard computer vision. Au-

tonomous Robots, 33(1-2):21–39, 2012.

[46] N. Michael, D. Mellinger, Q. Lindsey, and V. Kumar. The grasp multiple micro-uav

testbed. Robotics & Automation Magazine, IEEE, 17(3):56–65, 2010.

[47] T. Mikami and K. Uchiyama. Design of flight control system for quad tilt-wing uav. In

Unmanned Aircraft Systems (ICUAS), 2015 International Conference on, pages 801–805.

IEEE, 2015.

[48] M. W. Mueller and R. D’Andrea. Stability and control of a quadrocopter despite the

108
 complete loss of one, two, or three propellers. In Robotics and Automation (ICRA), 2014

IEEE International Conference on, pages 45–52. IEEE, 2014.

[49] D. R. Nelson, D. B. Barber, T. W. McLain, and R. W. Beard. Vector field path following

for miniature air vehicles. Robotics, IEEE Transactions on, 23(3):519–529, 2007.

[50] A. Nemati, M. HZarif, and M. M. Fateh. Helicopter adaptive control with parameter es-

timation based on feedback linearization. International Journal of Mechanical Industrial

and Aerospace Engineering, 2:229–234, 2007.

[51] A. Nemati and M. Kumar. Modeling and control of a single axis tilting quadcopter. In

American Control Conference (ACC), 2014, pages 3077–3082. IEEE, 2014.

[52] A. Nemati and M. Kumar. Non-linear control of tilting-quadcopter using feedback lin-

earization based motion control. In ASME 2014 Dynamic Systems and Control Confer-

ence, pages V003T48A005–V003T48A005. American Society of Mechanical Engineers,

2014.

[53] A. Nemati and M. Kumar. Control of microcoaxial helicopter based on a reduced-order

observer. Journal of Aerospace Engineering, page 04015074, 2015.

[54] A. Nemati, R. Kumar, and M. Kumar. Stabilizing and control of tilting-rotor quadcopter

in case of a propeller failure. In ASME 2016 Dynamic systems and control conference.

American Society of Mechanical Engineers, 2016.

[55] A. Nemati, N. Soni, M. Sarim, and M. Kumar. Design, fabrication and control of a tilt

rotor quadcopter: Theory and experiments. In ASME 2016 Dynamic systems and control

conference. American Society of Mechanical Engineers, 2016.

[56] K. Nonami, F. Kendoul, S. Suzuki, W. Wang, and D. Nakazawa. Autonomous Flying

Robots: Unmanned Aerial Vehicles and Micro Aerial Vehicles. Springer Science & Busi-

ness Media, 2010.

[57] B. Norton. Bell Boeing V-22 Osprey: tiltrotor tactical transport. Aerofax, 2004.

[58] K. Ogata. Modern control engineering. Prentice Hall PTR, 2001.

109
[59] K. T. Oner, E. Cetinsoy, M. Unel, M. F. Aksit, I. Kandemir, and K. Gulez. Dynamic

model and control of a new quadrotor unmanned aerial vehicle with tilt-wing mechanism.

International Journal of Applied Science, Engineering & Technology, 5(2), 2009.

[60] W. R. Oney. High power density brushless dc motor, Feb. 5 1980. US Patent 4,187,441.

[61] A. Oosedo, S. Abiko, S. Narasaki, A. Kuno, A. Konno, and M. Uchiyama. Flight con-

trol systems of a quad tilt rotor unmanned aerial vehicle for a large attitude change. In

Robotics and Automation (ICRA), 2015 IEEE International Conference on, pages 2326–

2331. IEEE, 2015.

[62] C. Papachristos, K. Alexis, and A. Tzes. Hybrid model predictive flight mode conversion

control of unmanned quad-tiltrotors. In Control Conference (ECC), 2013 European, pages

1793–1798. IEEE, 2013.

[63] P. Pounds and R. Mahony. Design principles of large quadrotors for practical applications.

In Robotics and Automation, 2009. ICRA’09. IEEE International Conference on, pages

3265–3270. IEEE, 2009.

[64] P. Pounds, R. Mahony, and P. Corke. Modelling and control of a quad-rotor robot. In Pro-

ceedings Australasian Conference on Robotics and Automation 2006. Australian Robotics

and Automation Association Inc., 2006.

[65] L. Qing, C. Zhihao, Y. Jinpeng, S. Yun, and W. YingXun. Trajectory tracking control for

hovering and acceleration maneuver of quad tilt rotor uav. In Control Conference (CCC),

2014 33rd Chinese, pages 2052–2057. IEEE, 2014.

[66] M. Quigley, K. Conley, B. Gerkey, J. Faust, T. Foote, J. Leibs, R. Wheeler, and A. Y. Ng.

Ros: an open-source robot operating system. In ICRA workshop on open source software,

volume 3, page 5, 2009.

[67] M. Radmanesh and M. Kumar. Grey wolf optimization based sense and avoid algorithm

for UAV path planning in uncertain environment using a bayesian framework. In 2016

International Conference on Unmanned Aircraft Systems (ICUAS), pages 68–76. IEEE,

2016.
110
[68] M. Radmanesh, M. Kumar, A. Nemati, and M. Sarim. Dynamic optimal UAV trajectory

planning in the national airspace system via mixed integer linear programming. Proceed-

ings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineer-

ing, page 0954410015609361, 2015.

[69] M. Radmanesh, M. Kumar, A. Nemati, and M. Sarim. Solution of traveling salesman

problem with hotel selection in the framework of MILP-Tropical optimization. In 2016

American Control Conference. IEEE, 2016.

[70] M. Radmanesh, M. Kumar, and M. Sarim. On the effect of different splines on way-point

navigation of quad-copters. In ASME 2016 Dynamic systems and control conference.

American Society of Mechanical Engineers, 2016.

[71] M. Radmanesh, A. Nemati, M. Sarim, and M. Kumar. Flight formation of quad-copters in

presence of dynamic obstacles using mixed integer linear programming. In ASME 2015

Dynamic systems and control conference, pages V001T06A009–V001T06A009. Ameri-

can Society of Mechanical Engineers, 2015.

[72] M. Radmanesh and I. Samani. IUT MAV2013, Part I: Aerodynamic design of tailless

wing and body configuration.

[73] M. Radmanesh and I. Samani. IUT MAV2013, Part II: flight test results.

[74] A. Rodić and G. Mester. The modeling and simulation of an autonomous quad-rotor

microcopter in a virtual outdoor scenario. Acta Polytechnica Hungarica, 8(4):107–122,

2011.

[75] J. Roldan, D. Sanz, J. del Cerro, and A. Barrientos. Lift failure detection and management

system for quadrotors. In ROBOT2013: First Iberian Robotics Conference, pages 103–

114. Springer, 2014.

[76] M. Ryll, H. Bulthoff, and P. R. Giordano. Modeling and control of a quadrotor uav

with tilting propellers. In Robotics and Automation (ICRA), 2012 IEEE International

Conference on, pages 4606–4613. IEEE, 2012.

111
[77] M. Ryll, H. H. Bulthoff, and P. R. Giordano. First flight tests for a quadrotor uav with tilt-

ing propellers. In Robotics and Automation (ICRA), 2013 IEEE International Conference

on, pages 295–302. IEEE, 2013.

[78] M. Ryll, H. H. Bülthoff, and P. R. Giordano. A novel overactuated quadrotor unmanned

aerial vehicle: modeling, control, and experimental validation. IEEE Transactions on

Control Systems Technology, 23(2):540–556, 2015.

[79] E. Samiei, M. Nazari, E. Butcher, and H. Schaub. Delayed feedback control of rigid

body attitude using neural networks and lyapunov–krasovskii functionals. In AAS/AIAA

Spaceflight Mechanics Meeting, Charleston, SC, pages 12–168, 2012.

[80] A. Sanyal, M. Izadi, G. Misra, E. Samiei, and D. Scheeres. Estimation of dynamics of

space objects from visual feedback during proximity operations. In AIAA/AAS Astrody-

namics Specialist Conference, AIAA, volume 4419, pages 4–7, 2014.

[81] P. Segui-Gasco, Y. Al-Rihani, H.-S. Shin, and A. Savvaris. A novel actuation concept for

a multi rotor uav. Journal of Intelligent & Robotic Systems, 74(1-2):173–191, 2014.

[82] F. Senkul and E. Altug. Modeling and control of a novel tilt-roll rotor quadrotor uav.

In Unmanned Aircraft Systems (ICUAS), 2013 International Conference on, pages 1071–

1076. IEEE, 2013.

[83] S. Sepasi, R. Ghorbani, and B. Y. Liaw. Soc estimation for aged lithium-ion batteries using

model adaptive extended kalman filter. In Transportation Electrification Conference and

Expo (ITEC), 2013 IEEE, pages 1–6. IEEE, 2013.

[84] S. Sepasi, R. Ghorbani, and B. Y. Liaw. Improved extended kalman filter for state of

charge estimation of battery pack. Journal of Power Sources, 255:368–376, 2014.

[85] S. Shafiei and S. Sepasi. Incorporating sliding mode and fuzzy controller with bounded

torques for set-point tracking of robot manipulators. Elektronika ir elektrotechnika,

104(8):3–8, 2015.

112
[86] J.-J. E. Slotine, W. Li, et al. Applied nonlinear control, volume 199. Prentice-Hall Engle-

wood Cliffs, NJ, 1991.

[87] S. H. Strogatz. Nonlinear dynamics and chaos: with applications to physics, biology,

chemistry, and engineering. Westview press, 2014.

[88] E. Stura and C. Nicolini. New nanomaterials for light weight lithium batteries. Analytica

chimica acta, 568(1):57–64, 2006.

[89] A. Tayebi and S. McGilvray. Attitude stabilization of a vtol quadrotor aircraft. IEEE

Transactions on control systems technology, 14(3):562–571, 2006.

[90] H. P. Thien, T. Mulyanto, H. Muhammad, and S. Suzuki. Mathematical modeling and

experimental identification of micro coaxial helicopter dynamics. International Journal

of Basic & Applied Sciences, 12(2), 2012.

[91] M. N. Thompson. Flight direction control system for blimps, May 25 1999. US Patent

5,906,335.

[92] L. A. Zadeh and C. A. Deoser. Linear system theory. Robert E. Krieger Publishing

Company, 1976.

[93] Y. Zhang and J. Jiang. Bibliographical review on reconfigurable fault-tolerant control

systems. Annual reviews in control, 32(2):229–252, 2008.

[94] Q.-L. Zhou, Y. Zhang, C.-A. Rabbath, and D. Theilliol. Design of feedback lineariza-

tion control and reconfigurable control allocation with application to a quadrotor uav. In

Control and Fault-Tolerant Systems (SysTol), 2010 Conference on, pages 371–376. IEEE,

2010.

113